SlideShare a Scribd company logo
17.08.2020 Concrete compressive strength
212.3.101.119:3890 1/34
Concrete compressive strength
We have a numerical dataset with information about di erent batches of concrete. Each batch
is Strength of Concrete in speci c moment in time (number of days after prodaction). It is
known which and how much components are in of Concrete. There are information about
1030 batches in the dataset.
Cement Slag Ash Water Plasticizer Breakstone Sand Time Strength
540.0 0.0 0 162 2.5 1040.0 676.0 28 79.99
540.0 0.0 0 162 2.5 1055.0 676.0 28 61.89
332.5 142.5 0 228 0.0 932.0 594.0 270 40.27
332.5 142.5 0 228 0.0 932.0 594.0 365 41.05
198.6 132.4 0 192 0.0 978.4 825.5 360 44.30
Main components for production: Cement, Water, Sand, Breakstone. These components are
always using for Concrete production.
Additional components: Not all batches contain Slag, Ash or Plasticizer. Plasticizer is used to
add less water. If we add Ash, we can add less Cement. Furnace Slag in uences on the concrete
structure.
Time: There are bathes with the same composition of components, but with di erent number
of days. The Concrete Strength has been checked after 1,3, 7, 14, 28, 56, 90, 91, 100, 120, 180,
270, 360, 365 days after prodaction.There are a few data observations where Time is 1, 120
days in the dataset. We have exclude this observation from consideration. We have little data
with Time = 90, 91, 100 and Time = 270, 360, 365. We use 90% of all batches.
When, its strength has reached 99% in 28 days, still concrete continues to gain strength after
that period, but that rate of gain in compressive strength is very less compared to that in 28
days. So, data with Time = 90, 91 we have set Time = 90 and observations with Time = 270, 360,
365 we have set Time = 365. So, there are 1030 observations in the dataset, we use 1025.
1m3
1m3
17.08.2020 Concrete compressive strength
212.3.101.119:3890 2/34
Our task: determine how a change in the proportions of di erent elements a ects the
commpresive strength of concrete.
Base analysis
Base vizualization
We can see strong dependence between Cement and Strength. The positive value of the slope
in the Linear Model (Strength~Cement) shows that If we add more Cement, Concrete Strength
increases. The negative values of the slope in the Linear Models (Strength~Water,
Strength~Sand) show that It is inverse dependence between Strength and Water; Strength and
Sand. The value of the slope in the Breakstone~Strength less than other, it shows that
Breakstone practically doesn’t a ect on Strength.
17.08.2020 Concrete compressive strength
212.3.101.119:3890 3/34
We can see inverse dependence between Ash and Strength (the negative value of the slope).
The value of the slope in the Slag~Strength less than other, it shows that It is low dependence
between Slag and Strength. The positive value of the slope in the Plasticizer~Strength model
indicates that an increase in quantity Plasticizer improves Strength.
Correlation matrix
It shows that there isn’t big intercorrelation (>0.7) between explanatory variables and we don’t
need to delete any variables. It is easy to see that the most in uential components are Time
and Cement.We can see strong intercorrelation between Water and Plasticizer. If we add
Plasticizer, we can use less Water.
Before building a model we have checked for normality the dependent variable (Strength). This
Strength doesn’t Normally Distributed, but for Linear Regression this is an optional condition.
Building models
Hypothesis 1: Concrete compressive strength depends on all ingredients and Time.
##
17.08.2020 Concrete compressive strength
212.3.101.119:3890 4/34
##
## Call:
## lm(formula = Strength ~ ., data = data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -28.970 -6.455 0.403 6.875 34.255
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -25.686533 27.145205 -0.946 0.344239
## Cement 0.120191 0.008657 13.883 < 2e-16 ***
## Slag 0.103565 0.010326 10.029 < 2e-16 ***
## Ash 0.088255 0.012815 6.887 9.99e-12 ***
## Water -0.146563 0.040955 -3.579 0.000362 ***
## Plasticizer 0.291271 0.094866 3.070 0.002195 **
## Breakstone 0.019292 0.009598 2.010 0.044692 *
## Sand 0.021507 0.010921 1.969 0.049199 *
## Time 0.101555 0.005176 19.620 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 10.55 on 1016 degrees of freedom
## Multiple R-squared: 0.6043, Adjusted R-squared: 0.6012
## F-statistic: 193.9 on 8 and 1016 DF, p-value: < 2.2e-16
Cement, Slag, Ash, Water, Time, Breakstone and Sand strongly a ect Strength. The R-squared is
0.6012, it indicates that 60% of the variation in the Strength is explained by the input variables.
Model 1 check. Conditionals of Gauss–Markov theorem.
1)Homoscedasticity of the dispersion of residues.
p = 0 < 0.05, it means that residual variance is related to model predicted values.Failed
3) An auto-correlation check ( The Durbin–Watson statistic).
p = 0 < 0.05, it means that it is auto-correlation. Failed
4)The normal distribution of residues (The Skewness-Kurtosis’s Test).
17.08.2020 Concrete compressive strength
212.3.101.119:3890 5/34
Residues are distributed normally. Sucess
17.08.2020 Concrete compressive strength
212.3.101.119:3890 6/34
Conclusion: Not all conditions of the Gauss–Markov theorem are satis ed. Our hypothesis is
false, so Concrete compressive strength doesn’t depend on all ingredients and Time and we
can’t use our model.
We have made some changes: built Linear Model (Strength ~log(Time)), using regression
coe cients, counted predicted values of Strength and after subtraction predicted Strength
from primary Strength, we have got values Strength without Time.
17.08.2020 Concrete compressive strength
212.3.101.119:3890 7/34
Hypothesis 2: Concrete compressive strength depends on all ingredients, besides Time.
##
## Call:
## lm(formula = strengthWithoutTime ~ Slag + Ash + Plasticizer +
## Breakstone + Sand + Cement + Water, data = data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -23.7942 -4.4161 -0.5019 3.9322 29.9414
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -84.013818 18.711087 -4.490 7.93e-06 ***
## Slag 0.113116 0.007079 15.978 < 2e-16 ***
## Ash 0.092166 0.008777 10.500 < 2e-16 ***
## Plasticizer 0.147213 0.065461 2.249 0.0247 *
## Breakstone 0.029070 0.006616 4.394 1.23e-05 ***
## Sand 0.033932 0.007501 4.524 6.79e-06 ***
## Cement 0.133024 0.005965 22.301 < 2e-16 ***
## Water -0.122236 0.028234 -4.329 1.64e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 7.283 on 1017 degrees of freedom
## Multiple R-squared: 0.7301, Adjusted R-squared: 0.7282
## F-statistic: 392.9 on 7 and 1017 DF, p-value: < 2.2e-16
Cement, Slag, Ash, Water, Time, Breakstone and Sand strongly a ect Strength.The R-squared is
0.7282, it indicates that 72% of the variation in the Strength is explained by the input variables.
Model 2 check. Conditionals of Gauss–Markov theorem
17.08.2020 Concrete compressive strength
212.3.101.119:3890 8/34
1)Homoscedasticity of the dispersion of residues.
p = 0 < 0.05, it means that residual variance is related to model predicted values. Failed
2) An auto-correlation check ( The Durbin–Watson statistic).
p = 0 < 0.05, it means that it is auto-correlation. Failed
3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
Residues don’t have the normal distribution. Failed
17.08.2020 Concrete compressive strength
212.3.101.119:3890 9/34
Conclusion: Our hypothesis is false. Model Strength~Time is wrong, because Conditionals of
Gauss–Markov theorem are failed.
Let’s divide the dataset into two parts: rst part includes observations with main and additional
components; second part includes observations only with main components. We have build
two models for each sets.
17.08.2020 Concrete compressive strength
212.3.101.119:3890 10/34
Hypothesis 3: Concrete compressive strength depends on only Main components.
##
## Call:
## lm(formula = strengthWithoutTime ~ Cement + Water + Breakstone +
## Sand, data = data_base)
##
## Residuals:
## Min 1Q Median 3Q Max
## -13.0673 -2.8142 -0.0592 2.0592 19.8428
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 76.471423 28.553201 2.678 0.00802 **
## Cement 0.099861 0.009644 10.355 < 2e-16 ***
## Water -0.348389 0.047123 -7.393 3.9e-12 ***
## Breakstone -0.023489 0.008764 -2.680 0.00798 **
## Sand -0.027102 0.012582 -2.154 0.03244 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 4.796 on 199 degrees of freedom
## Multiple R-squared: 0.8323, Adjusted R-squared: 0.8289
## F-statistic: 246.9 on 4 and 199 DF, p-value: < 2.2e-16
Cement, Slag, Ash, Water, Breakstone and Sand strongly a ect Strength, especially Cement and
Water.The R-squared is 0.8289, it indicates that 82% of the variation in the Strength is explained
by the input variables.
Model 3 check. Conditionals of Gauss–Markov theorem
1)Homoscedasticity of the dispersion of residues.
p = 0.0004 < 0.05, it means that residual variance is related to model predicted values. Failed
2) An auto-correlation check ( The Durbin–Watson statistic).
p = 0.016 < 0.05, it means that it is auto-correlation. Failed
3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
17.08.2020 Concrete compressive strength
212.3.101.119:3890 11/34
Residues don’t have the normal distribution. Failed
17.08.2020 Concrete compressive strength
212.3.101.119:3890 12/34
Conclusion: Our hypothesis is false. We can’t use model 3. Model Strength~Time is wrong,
because Conditionals of Gauss–Markov theorem are failed.
Hypothesis 4: Concrete compressive strength depends on all components (There are
additional components in all observations).
##
## Call:
## lm(formula = strengthWithoutTime ~ Cement + Slag + Ash + Water +
## Plasticizer + Breakstone + Sand, data = data_all)
##
## Residuals:
## Min 1Q Median 3Q Max
## -23.210 -4.804 -0.552 4.169 28.024
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -1.216e+02 2.181e+01 -5.578 3.32e-08 ***
## Cement 1.433e-01 6.953e-03 20.618 < 2e-16 ***
## Slag 1.130e-01 8.595e-03 13.149 < 2e-16 ***
## Ash 9.070e-02 1.046e-02 8.675 < 2e-16 ***
## Water -7.444e-02 3.202e-02 -2.324 0.0203 *
## Plasticizer 9.944e-02 7.315e-02 1.359 0.1744
## Breakstone 4.513e-02 7.919e-03 5.699 1.68e-08 ***
## Sand 4.719e-02 8.473e-03 5.570 3.47e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 7.357 on 813 degrees of freedom
## Multiple R-squared: 0.7278, Adjusted R-squared: 0.7255
## F-statistic: 310.6 on 7 and 813 DF, p-value: < 2.2e-16
17.08.2020 Concrete compressive strength
212.3.101.119:3890 13/34
Cement, Slag, Ash, Water, Breakstone and Sand strongly a ect Strength.The R-squared is
0.7255, it indicates that 72% of the variation in the Strength is explained by the input variables.
Model 4 check. Conditionals of Gauss–Markov theorem
1)Homoscedasticity of the dispersion of residues.
p = 0 < 0.05, it means that residual variance is related to model predicted values.Failed
2) An auto-correlation check ( The Durbin–Watson statistic).
p = 0 < 0.05, it means that it is auto-correlation. Failed
3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
Residues don’t have the normal distribution. Failed
17.08.2020 Concrete compressive strength
212.3.101.119:3890 14/34
Conclusion: Our hypothesis is false. We can’t use model 4. Model Strength~Time is wrong,
because Conditionals of Gauss–Markov theorem are failed.
All our models are wrong, so we have to apply a di erent analysis approach. We found out and
deleted dublicate data (with the same compositon, Time and Strength) and wrong batches. We
have got 988 batches. After that, we have grouped batches with di erent composotion, but the
same Time. As a result, we have got groups with: Time = 3, 7, 14, 28. Batches with Time = 180,
17.08.2020 Concrete compressive strength
212.3.101.119:3890 15/34
270, 360, 365 were united in one group with Time = 300;with Time = 90, 91, 100 were united in
one group with Time = 95, because we have found approximately the same values of Strength
for batches with Time = 180, 270, 360, 365 and for Time = 90, 91, 100.
We have created 6 datasets with 125, 117, 62, 417, 86, 122, 59 observations, where Time is
respectively 3, 7, 14, 28, 56, 95 (90, 91, 100), 300 (180, 270, 360, 365).
Time NumberOfBatches
3 125
7 117
14 62
28 417
56 86
95 122
300 59
At rst, we have checked the normal distribution of Strength for di erent groups. It was
successful for group with Time = 56, 95, 300, it means that in these datasets Stregth has the
normal distribution. But, we will build model for all datasets, because it’s uncritical for Linear
Model that predicted variable doesn’t have the normal distribution.
Time = 3 Hypothesis: Concrete compressive strength depends on all components.
##
## Call:
## lm(formula = Strength ~ Cement + Sand + Breakstone + Ash + Slag,
## data = d3)
##
17.08.2020 Concrete compressive strength
212.3.101.119:3890 16/34
## Residuals:
## Min 1Q Median 3Q Max
## -9.3248 -1.9449 -0.0407 2.2557 8.6397
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -1.398e+02 1.341e+01 -10.422 < 2e-16 ***
## Cement 1.268e-01 5.281e-03 24.015 < 2e-16 ***
## Sand 6.790e-02 7.167e-03 9.474 3.30e-16 ***
## Breakstone 5.815e-02 7.301e-03 7.964 1.11e-12 ***
## Ash 9.115e-02 7.446e-03 12.241 < 2e-16 ***
## Slag 9.537e-02 7.130e-03 13.377 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.701 on 119 degrees of freedom
## Multiple R-squared: 0.859, Adjusted R-squared: 0.8531
## F-statistic: 145 on 5 and 119 DF, p-value: < 2.2e-16
Cement, Slag, Ash, Breakstone and Sand strongly a ect Strength.The R-squared is 0.8531, it
indicates that 85% of the variation in the Strength is explained by the input variables.
Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues.
p = 0 < 0.05, it means that residual variance is related to model predicted values.Failed
2) An auto-correlation check ( The Durbin–Watson statistic).
p = 0.02,it means that it isn’t auto-correlation. Sucess
3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
Residues have the normal distribution. Sucess
17.08.2020 Concrete compressive strength
212.3.101.119:3890 17/34
Conclusion: Our hypothesis is false. Because one contitional of Gauss–Markov theorem is
failed.
Time = 7 Hypothesis: Concrete compressive strength depends on all components.
##
## Call:
17.08.2020 Concrete compressive strength
212.3.101.119:3890 18/34
## Call:
## lm(formula = Strength ~ Cement + Water + Plasticizer + Slag -
## 1, data = d7)
##
## Residuals:
## Min 1Q Median 3Q Max
## -15.0102 -2.8381 -0.4443 2.5889 13.4414
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## Cement 0.105573 0.005572 18.947 < 2e-16 ***
## Water -0.076393 0.010230 -7.468 1.84e-11 ***
## Plasticizer 0.590486 0.091078 6.483 2.44e-09 ***
## Slag 0.052515 0.005285 9.936 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 4.614 on 113 degrees of freedom
## Multiple R-squared: 0.9757, Adjusted R-squared: 0.9749
## F-statistic: 1137 on 4 and 113 DF, p-value: < 2.2e-16
Cement, Slag, Water, Plasticizer strongly a ect Strength.The R-squared is 0.9749, it indicates
that 97.5% of the variation in the Strength is explained by the input variables.
Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues.
p = 0 < 0.05, it means that residual variance is related to model predicted values.Failed
2) An auto-correlation check ( The Durbin–Watson statistic).
p = 0.378,it means that it isn’t auto-correlation. Sucess
3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
17.08.2020 Concrete compressive strength
212.3.101.119:3890 19/34
Residues have the normal distribution. Sucess
17.08.2020 Concrete compressive strength
212.3.101.119:3890 20/34
Conclusion: Our hypothesis is false. Because Homoscedasticity of the dispersion of residues is
failed.
Time = 14 Hypothesis: Concrete compressive strength depends on all components.
##
## Call:
## lm(formula = Strength ~ Cement + Water + Sand + Plasticizer +
## Ash + Slag, data = d14)
##
## Residuals:
## Min 1Q Median 3Q Max
## -11.9751 -2.4389 -0.3798 2.8660 7.9254
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -42.39184 20.43416 -2.075 0.04272 *
## Cement 0.16392 0.01685 9.726 1.50e-13 ***
## Water -0.09603 0.04460 -2.153 0.03570 *
## Sand 0.03811 0.01422 2.680 0.00968 **
## Plasticizer 0.60109 0.20272 2.965 0.00447 **
## Ash 0.10555 0.02217 4.760 1.45e-05 ***
## Slag 0.14019 0.02108 6.650 1.41e-08 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 4.03 on 55 degrees of freedom
## Multiple R-squared: 0.8038, Adjusted R-squared: 0.7824
## F-statistic: 37.55 on 6 and 55 DF, p-value: < 2.2e-16
17.08.2020 Concrete compressive strength
212.3.101.119:3890 21/34
Cement, Water, Slag, Ash, Plasticizer, Sand strongly a ect Strength.The R-squared is 0.7824, it
indicates that 78.2% of the variation in the Strength is explained by the input variables.
Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues.
p = 0.051289, it means that residual variance isn’t related to model predicted values.Sucess
2) An auto-correlation check ( The Durbin–Watson statistic).
p = 0.742,it means that it isn’t auto-correlation. Sucess
3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
Residues have the normal distribution. Sucess
17.08.2020 Concrete compressive strength
212.3.101.119:3890 22/34
Conclusion: Our hypothesis is true. It means that Least squares is better for this data and we
can use this model.
Time = 28 Hypothesis: Concrete compressive strength depends on all components.
##
## Call:
17.08.2020 Concrete compressive strength
212.3.101.119:3890 23/34
## Call:
## lm(formula = Strength ~ Cement + Water + Sand + Breakstone +
## Ash + Slag, data = d28)
##
## Residuals:
## Min 1Q Median 3Q Max
## -20.646 -4.046 -1.357 3.436 27.835
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) -79.647940 27.560586 -2.890 0.00406 **
## Cement 0.165211 0.009482 17.423 < 2e-16 ***
## Water -0.090868 0.035966 -2.526 0.01190 *
## Sand 0.049274 0.011697 4.213 3.11e-05 ***
## Breakstone 0.034402 0.009934 3.463 0.00059 ***
## Ash 0.104380 0.014159 7.372 9.36e-13 ***
## Slag 0.139517 0.011267 12.383 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 7.104 on 410 degrees of freedom
## Multiple R-squared: 0.7609, Adjusted R-squared: 0.7574
## F-statistic: 217.4 on 6 and 410 DF, p-value: < 2.2e-16
Cement, Slag, Water, Plasticizer, Sand, Breakstone, Ash strongly a ect Strength.The R-squared
is 0.7574, it indicates that 75% of the variation in the Strength is explained by the input
variables.
Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues.
p = 0 < 0.05, it means that residual variance is related to model predicted values.Failed
2) An auto-correlation check ( The Durbin–Watson statistic).
p = 0,it means that it is auto-correlation. Failed
3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
17.08.2020 Concrete compressive strength
212.3.101.119:3890 24/34
Residues don’t have the normal distribution. Failed
17.08.2020 Concrete compressive strength
212.3.101.119:3890 25/34
Conclusion: Our hypothesis is false. All condidtionals of Gauss–Markov theorem are failed. We
have to split into several parts this dataset and build the model again.
Time = 56 Hypothesis: Concrete compressive strength depends on all components.
##
## Call:
## lm(formula = Strength ~ Cement + Water + Sand + Ash + Slag -
## 1, data = d56)
##
## Residuals:
## Min 1Q Median 3Q Max
## -18.7469 -4.0671 -0.8282 4.7413 16.7102
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## Cement 0.103363 0.008088 12.779 < 2e-16 ***
## Water -0.140737 0.029632 -4.749 8.69e-06 ***
## Sand 0.032485 0.005855 5.548 3.53e-07 ***
## Ash 0.117464 0.017629 6.663 2.99e-09 ***
## Slag 0.161311 0.012577 12.826 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 6.468 on 81 degrees of freedom
## Multiple R-squared: 0.9857, Adjusted R-squared: 0.9848
## F-statistic: 1118 on 5 and 81 DF, p-value: < 2.2e-16
Cement, Water, Slag, Ash, Sand strongly a ect Strength.The R-squared is 0.9848, it indicates
that 98% of the variation in the Strength is explained by the input variables.
Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues.
17.08.2020 Concrete compressive strength
212.3.101.119:3890 26/34
p = 0.57373, it means that residual variance isn’t related to model predicted values.Sucess
2) An auto-correlation check ( The Durbin–Watson statistic).
p = 0.11,it means that it isn’t auto-correlation. Sucess
3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
Residues have the normal distribution. Sucess
17.08.2020 Concrete compressive strength
212.3.101.119:3890 27/34
Conclusion: Our hypothesis is true. It means that Least squares is better for this data and we
can use this model.
Time = 95 Hypothesis: Concrete compressive strength depends on all components.
##
## Call:
17.08.2020 Concrete compressive strength
212.3.101.119:3890 28/34
## Call:
## lm(formula = Strength ~ Cement + Water + Ash + Slag, data = d95)
##
## Residuals:
## Min 1Q Median 3Q Max
## -18.8475 -3.0299 -0.7854 3.3528 16.1291
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 40.938765 7.004618 5.845 4.68e-08 ***
## Cement 0.110679 0.007575 14.611 < 2e-16 ***
## Water -0.202701 0.028434 -7.129 9.00e-11 ***
## Ash 0.113753 0.015122 7.522 1.20e-11 ***
## Slag 0.109286 0.008733 12.514 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 6.071 on 117 degrees of freedom
## Multiple R-squared: 0.7829, Adjusted R-squared: 0.7755
## F-statistic: 105.5 on 4 and 117 DF, p-value: < 2.2e-16
Cement, Water, Slag, Ash strongly a ect Strength.The R-squared is 0.7755, it indicates that 77%
of the variation in the Strength is explained by the input variables.
Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues.
p = 0.067624, it means that residual variance isn’t related to model predicted values.Sucess
2) An auto-correlation check ( The Durbin–Watson statistic).
p = 0.476,it means that it isn’t auto-correlation. Sucess
3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
17.08.2020 Concrete compressive strength
212.3.101.119:3890 29/34
Residues have the normal distribution. Sucess
17.08.2020 Concrete compressive strength
212.3.101.119:3890 30/34
Conclusion: Our hypothesis is true. It means that Least squares is better for this data and we
can use this model.
Time = 300 Hypothesis: Concrete compressive strength depends on all components.
##
## Call:
## lm(formula = Strength ~ Cement + Sand + Breakstone + Slag, data = d300)
##
## Residuals:
## Min 1Q Median 3Q Max
## -14.974 -4.686 -1.429 6.205 14.222
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 89.06778 31.56549 2.822 0.00667 **
## Cement -0.15165 0.05226 -2.902 0.00536 **
## Sand -0.17691 0.04058 -4.360 5.89e-05 ***
## Breakstone 0.14330 0.02781 5.152 3.73e-06 ***
## Slag -0.15096 0.05292 -2.852 0.00614 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 7.113 on 54 degrees of freedom
## Multiple R-squared: 0.5813, Adjusted R-squared: 0.5503
## F-statistic: 18.74 on 4 and 54 DF, p-value: 1.034e-09
Cement, Water, Slag, Ash strongly a ect Strength.The R-squared is 0.5503 , it indicates that 55%
of the variation in the Strength is explained by the input variables.
Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues.
17.08.2020 Concrete compressive strength
212.3.101.119:3890 31/34
p = 0.13509, it means that residual variance isn’t related to model predicted values.Sucess
2) An auto-correlation check ( The Durbin–Watson statistic).
p = 0,it means that it is auto-correlation. Failed
3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
Residues have the normal distribution. Sucess
17.08.2020 Concrete compressive strength
212.3.101.119:3890 32/34
Conclusion: Our hypothesis is false. Because An auto-correlation check is failed.
17.08.2020 Concrete compressive strength
212.3.101.119:3890 33/34
If we use model for Time = 14 for all data, we get such result.
17.08.2020 Concrete compressive strength
212.3.101.119:3890 34/34

More Related Content

Similar to Concrete compressive strength

Effect of Steel Fibre and Marble Dust on the Mechanical Properties of High St...
Effect of Steel Fibre and Marble Dust on the Mechanical Properties of High St...Effect of Steel Fibre and Marble Dust on the Mechanical Properties of High St...
Effect of Steel Fibre and Marble Dust on the Mechanical Properties of High St...
IRJET Journal
 
Comparative Study on Flexural Strength of M20 Grade Beam with Lapping of Bars
Comparative Study on Flexural Strength of M20 Grade Beam with Lapping of BarsComparative Study on Flexural Strength of M20 Grade Beam with Lapping of Bars
Comparative Study on Flexural Strength of M20 Grade Beam with Lapping of Bars
IRJET Journal
 
Experimental Investigation on Concrete with Partial Replacement of Fine Aggre...
Experimental Investigation on Concrete with Partial Replacement of Fine Aggre...Experimental Investigation on Concrete with Partial Replacement of Fine Aggre...
Experimental Investigation on Concrete with Partial Replacement of Fine Aggre...
IRJET Journal
 
Experimental behaviour and analysis of stress in rigid pavement
Experimental behaviour and analysis of stress in rigid pavementExperimental behaviour and analysis of stress in rigid pavement
Experimental behaviour and analysis of stress in rigid pavementVivek Loyola
 
Concrete Mix Design.pptx
Concrete Mix Design.pptxConcrete Mix Design.pptx
Concrete Mix Design.pptx
ADCET, Ashta
 
An experimental study on mud concrete using soil as a fine aggrgate and ld sl...
An experimental study on mud concrete using soil as a fine aggrgate and ld sl...An experimental study on mud concrete using soil as a fine aggrgate and ld sl...
An experimental study on mud concrete using soil as a fine aggrgate and ld sl...
eSAT Journals
 
Assessment of Properties of Concrete with Partial Replacement by Coconut Cons...
Assessment of Properties of Concrete with Partial Replacement by Coconut Cons...Assessment of Properties of Concrete with Partial Replacement by Coconut Cons...
Assessment of Properties of Concrete with Partial Replacement by Coconut Cons...
IRJET Journal
 
IRJET - Experimental and Analytical Study on Partial Replacement of Coarse Ag...
IRJET - Experimental and Analytical Study on Partial Replacement of Coarse Ag...IRJET - Experimental and Analytical Study on Partial Replacement of Coarse Ag...
IRJET - Experimental and Analytical Study on Partial Replacement of Coarse Ag...
IRJET Journal
 
8th sem PPT Major Project .pptx
8th sem PPT Major Project .pptx8th sem PPT Major Project .pptx
8th sem PPT Major Project .pptx
MonalishaMishra21CEL
 
Properities Of Plain & Slag Concrete
Properities Of Plain & Slag ConcreteProperities Of Plain & Slag Concrete
Properities Of Plain & Slag Concrete
PRAKASH KUMAR
 
Compressive Strength of Different Grades of SCC Mix With 0.5% Of PEG 400 Self...
Compressive Strength of Different Grades of SCC Mix With 0.5% Of PEG 400 Self...Compressive Strength of Different Grades of SCC Mix With 0.5% Of PEG 400 Self...
Compressive Strength of Different Grades of SCC Mix With 0.5% Of PEG 400 Self...
IRJET Journal
 
Concrete mix design
Concrete mix designConcrete mix design
Concrete mix design
Adel Ba Wazir
 
concrete mix design.ppt
concrete mix design.pptconcrete mix design.ppt
concrete mix design.ppt
MohdNizamShakimon
 
Effect of controlled temperature (27°c 42°c)
Effect of controlled temperature (27°c  42°c)Effect of controlled temperature (27°c  42°c)
Effect of controlled temperature (27°c 42°c)prjpublications
 
IRJET- Probabilistic Investigation of the Compressive Strength of Perviou...
IRJET-  	  Probabilistic Investigation of the Compressive Strength of Perviou...IRJET-  	  Probabilistic Investigation of the Compressive Strength of Perviou...
IRJET- Probabilistic Investigation of the Compressive Strength of Perviou...
IRJET Journal
 
IRJET- VStudy on Strength and Durability Properties of Concrete using Steel S...
IRJET- VStudy on Strength and Durability Properties of Concrete using Steel S...IRJET- VStudy on Strength and Durability Properties of Concrete using Steel S...
IRJET- VStudy on Strength and Durability Properties of Concrete using Steel S...
IRJET Journal
 
Effect on Compressive and Tensile Strength of Cement Concrete Road Pavements ...
Effect on Compressive and Tensile Strength of Cement Concrete Road Pavements ...Effect on Compressive and Tensile Strength of Cement Concrete Road Pavements ...
Effect on Compressive and Tensile Strength of Cement Concrete Road Pavements ...
IRJET Journal
 
Electronic system for testing tensile strength of foundry sand
Electronic system for testing tensile strength of foundry sandElectronic system for testing tensile strength of foundry sand
Electronic system for testing tensile strength of foundry sand
eSAT Publishing House
 
A Study on Strength Properties of Concrete Replacing Partially the Natural Co...
A Study on Strength Properties of Concrete Replacing Partially the Natural Co...A Study on Strength Properties of Concrete Replacing Partially the Natural Co...
A Study on Strength Properties of Concrete Replacing Partially the Natural Co...
IRJET Journal
 

Similar to Concrete compressive strength (20)

Effect of Steel Fibre and Marble Dust on the Mechanical Properties of High St...
Effect of Steel Fibre and Marble Dust on the Mechanical Properties of High St...Effect of Steel Fibre and Marble Dust on the Mechanical Properties of High St...
Effect of Steel Fibre and Marble Dust on the Mechanical Properties of High St...
 
Comparative Study on Flexural Strength of M20 Grade Beam with Lapping of Bars
Comparative Study on Flexural Strength of M20 Grade Beam with Lapping of BarsComparative Study on Flexural Strength of M20 Grade Beam with Lapping of Bars
Comparative Study on Flexural Strength of M20 Grade Beam with Lapping of Bars
 
Experimental Investigation on Concrete with Partial Replacement of Fine Aggre...
Experimental Investigation on Concrete with Partial Replacement of Fine Aggre...Experimental Investigation on Concrete with Partial Replacement of Fine Aggre...
Experimental Investigation on Concrete with Partial Replacement of Fine Aggre...
 
Experimental behaviour and analysis of stress in rigid pavement
Experimental behaviour and analysis of stress in rigid pavementExperimental behaviour and analysis of stress in rigid pavement
Experimental behaviour and analysis of stress in rigid pavement
 
Mutt_Wind_Tunnel_Results_v2
Mutt_Wind_Tunnel_Results_v2Mutt_Wind_Tunnel_Results_v2
Mutt_Wind_Tunnel_Results_v2
 
Concrete Mix Design.pptx
Concrete Mix Design.pptxConcrete Mix Design.pptx
Concrete Mix Design.pptx
 
An experimental study on mud concrete using soil as a fine aggrgate and ld sl...
An experimental study on mud concrete using soil as a fine aggrgate and ld sl...An experimental study on mud concrete using soil as a fine aggrgate and ld sl...
An experimental study on mud concrete using soil as a fine aggrgate and ld sl...
 
Assessment of Properties of Concrete with Partial Replacement by Coconut Cons...
Assessment of Properties of Concrete with Partial Replacement by Coconut Cons...Assessment of Properties of Concrete with Partial Replacement by Coconut Cons...
Assessment of Properties of Concrete with Partial Replacement by Coconut Cons...
 
IRJET - Experimental and Analytical Study on Partial Replacement of Coarse Ag...
IRJET - Experimental and Analytical Study on Partial Replacement of Coarse Ag...IRJET - Experimental and Analytical Study on Partial Replacement of Coarse Ag...
IRJET - Experimental and Analytical Study on Partial Replacement of Coarse Ag...
 
8th sem PPT Major Project .pptx
8th sem PPT Major Project .pptx8th sem PPT Major Project .pptx
8th sem PPT Major Project .pptx
 
Properities Of Plain & Slag Concrete
Properities Of Plain & Slag ConcreteProperities Of Plain & Slag Concrete
Properities Of Plain & Slag Concrete
 
Compressive Strength of Different Grades of SCC Mix With 0.5% Of PEG 400 Self...
Compressive Strength of Different Grades of SCC Mix With 0.5% Of PEG 400 Self...Compressive Strength of Different Grades of SCC Mix With 0.5% Of PEG 400 Self...
Compressive Strength of Different Grades of SCC Mix With 0.5% Of PEG 400 Self...
 
Concrete mix design
Concrete mix designConcrete mix design
Concrete mix design
 
concrete mix design.ppt
concrete mix design.pptconcrete mix design.ppt
concrete mix design.ppt
 
Effect of controlled temperature (27°c 42°c)
Effect of controlled temperature (27°c  42°c)Effect of controlled temperature (27°c  42°c)
Effect of controlled temperature (27°c 42°c)
 
IRJET- Probabilistic Investigation of the Compressive Strength of Perviou...
IRJET-  	  Probabilistic Investigation of the Compressive Strength of Perviou...IRJET-  	  Probabilistic Investigation of the Compressive Strength of Perviou...
IRJET- Probabilistic Investigation of the Compressive Strength of Perviou...
 
IRJET- VStudy on Strength and Durability Properties of Concrete using Steel S...
IRJET- VStudy on Strength and Durability Properties of Concrete using Steel S...IRJET- VStudy on Strength and Durability Properties of Concrete using Steel S...
IRJET- VStudy on Strength and Durability Properties of Concrete using Steel S...
 
Effect on Compressive and Tensile Strength of Cement Concrete Road Pavements ...
Effect on Compressive and Tensile Strength of Cement Concrete Road Pavements ...Effect on Compressive and Tensile Strength of Cement Concrete Road Pavements ...
Effect on Compressive and Tensile Strength of Cement Concrete Road Pavements ...
 
Electronic system for testing tensile strength of foundry sand
Electronic system for testing tensile strength of foundry sandElectronic system for testing tensile strength of foundry sand
Electronic system for testing tensile strength of foundry sand
 
A Study on Strength Properties of Concrete Replacing Partially the Natural Co...
A Study on Strength Properties of Concrete Replacing Partially the Natural Co...A Study on Strength Properties of Concrete Replacing Partially the Natural Co...
A Study on Strength Properties of Concrete Replacing Partially the Natural Co...
 

Recently uploaded

ikea_woodgreen_petscharity_dog-alogue_digital.pdf
ikea_woodgreen_petscharity_dog-alogue_digital.pdfikea_woodgreen_petscharity_dog-alogue_digital.pdf
ikea_woodgreen_petscharity_dog-alogue_digital.pdf
agatadrynko
 
Authentically Social Presented by Corey Perlman
Authentically Social Presented by Corey PerlmanAuthentically Social Presented by Corey Perlman
Authentically Social Presented by Corey Perlman
Corey Perlman, Social Media Speaker and Consultant
 
Search Disrupted Google’s Leaked Documents Rock the SEO World.pdf
Search Disrupted Google’s Leaked Documents Rock the SEO World.pdfSearch Disrupted Google’s Leaked Documents Rock the SEO World.pdf
Search Disrupted Google’s Leaked Documents Rock the SEO World.pdf
Arihant Webtech Pvt. Ltd
 
ModelingMarketingStrategiesMKS.CollumbiaUniversitypdf
ModelingMarketingStrategiesMKS.CollumbiaUniversitypdfModelingMarketingStrategiesMKS.CollumbiaUniversitypdf
ModelingMarketingStrategiesMKS.CollumbiaUniversitypdf
fisherameliaisabella
 
Bài tập - Tiếng anh 11 Global Success UNIT 1 - Bản HS.doc
Bài tập - Tiếng anh 11 Global Success UNIT 1 - Bản HS.docBài tập - Tiếng anh 11 Global Success UNIT 1 - Bản HS.doc
Bài tập - Tiếng anh 11 Global Success UNIT 1 - Bản HS.doc
daothibichhang1
 
The Influence of Marketing Strategy and Market Competition on Business Perfor...
The Influence of Marketing Strategy and Market Competition on Business Perfor...The Influence of Marketing Strategy and Market Competition on Business Perfor...
The Influence of Marketing Strategy and Market Competition on Business Perfor...
Adam Smith
 
Set off and carry forward of losses and assessment of individuals.pptx
Set off and carry forward of losses and assessment of individuals.pptxSet off and carry forward of losses and assessment of individuals.pptx
Set off and carry forward of losses and assessment of individuals.pptx
HARSHITHV26
 
Recruiting in the Digital Age: A Social Media Masterclass
Recruiting in the Digital Age: A Social Media MasterclassRecruiting in the Digital Age: A Social Media Masterclass
Recruiting in the Digital Age: A Social Media Masterclass
LuanWise
 
Project File Report BBA 6th semester.pdf
Project File Report BBA 6th semester.pdfProject File Report BBA 6th semester.pdf
Project File Report BBA 6th semester.pdf
RajPriye
 
ikea_woodgreen_petscharity_cat-alogue_digital.pdf
ikea_woodgreen_petscharity_cat-alogue_digital.pdfikea_woodgreen_petscharity_cat-alogue_digital.pdf
ikea_woodgreen_petscharity_cat-alogue_digital.pdf
agatadrynko
 
Cree_Rey_BrandIdentityKit.PDF_PersonalBd
Cree_Rey_BrandIdentityKit.PDF_PersonalBdCree_Rey_BrandIdentityKit.PDF_PersonalBd
Cree_Rey_BrandIdentityKit.PDF_PersonalBd
creerey
 
Brand Analysis for an artist named Struan
Brand Analysis for an artist named StruanBrand Analysis for an artist named Struan
Brand Analysis for an artist named Struan
sarahvanessa51503
 
Mastering B2B Payments Webinar from BlueSnap
Mastering B2B Payments Webinar from BlueSnapMastering B2B Payments Webinar from BlueSnap
Mastering B2B Payments Webinar from BlueSnap
Norma Mushkat Gaffin
 
Putting the SPARK into Virtual Training.pptx
Putting the SPARK into Virtual Training.pptxPutting the SPARK into Virtual Training.pptx
Putting the SPARK into Virtual Training.pptx
Cynthia Clay
 
Call 8867766396 Satta Matka Dpboss Matka Guessing Satta batta Matka 420 Satta...
Call 8867766396 Satta Matka Dpboss Matka Guessing Satta batta Matka 420 Satta...Call 8867766396 Satta Matka Dpboss Matka Guessing Satta batta Matka 420 Satta...
Call 8867766396 Satta Matka Dpboss Matka Guessing Satta batta Matka 420 Satta...
bosssp10
 
Maksym Vyshnivetskyi: PMO Quality Management (UA)
Maksym Vyshnivetskyi: PMO Quality Management (UA)Maksym Vyshnivetskyi: PMO Quality Management (UA)
Maksym Vyshnivetskyi: PMO Quality Management (UA)
Lviv Startup Club
 
Evgen Osmak: Methods of key project parameters estimation: from the shaman-in...
Evgen Osmak: Methods of key project parameters estimation: from the shaman-in...Evgen Osmak: Methods of key project parameters estimation: from the shaman-in...
Evgen Osmak: Methods of key project parameters estimation: from the shaman-in...
Lviv Startup Club
 
BeMetals Investor Presentation_June 1, 2024.pdf
BeMetals Investor Presentation_June 1, 2024.pdfBeMetals Investor Presentation_June 1, 2024.pdf
BeMetals Investor Presentation_June 1, 2024.pdf
DerekIwanaka1
 
FINAL PRESENTATION.pptx12143241324134134
FINAL PRESENTATION.pptx12143241324134134FINAL PRESENTATION.pptx12143241324134134
FINAL PRESENTATION.pptx12143241324134134
LR1709MUSIC
 
In the Adani-Hindenburg case, what is SEBI investigating.pptx
In the Adani-Hindenburg case, what is SEBI investigating.pptxIn the Adani-Hindenburg case, what is SEBI investigating.pptx
In the Adani-Hindenburg case, what is SEBI investigating.pptx
Adani case
 

Recently uploaded (20)

ikea_woodgreen_petscharity_dog-alogue_digital.pdf
ikea_woodgreen_petscharity_dog-alogue_digital.pdfikea_woodgreen_petscharity_dog-alogue_digital.pdf
ikea_woodgreen_petscharity_dog-alogue_digital.pdf
 
Authentically Social Presented by Corey Perlman
Authentically Social Presented by Corey PerlmanAuthentically Social Presented by Corey Perlman
Authentically Social Presented by Corey Perlman
 
Search Disrupted Google’s Leaked Documents Rock the SEO World.pdf
Search Disrupted Google’s Leaked Documents Rock the SEO World.pdfSearch Disrupted Google’s Leaked Documents Rock the SEO World.pdf
Search Disrupted Google’s Leaked Documents Rock the SEO World.pdf
 
ModelingMarketingStrategiesMKS.CollumbiaUniversitypdf
ModelingMarketingStrategiesMKS.CollumbiaUniversitypdfModelingMarketingStrategiesMKS.CollumbiaUniversitypdf
ModelingMarketingStrategiesMKS.CollumbiaUniversitypdf
 
Bài tập - Tiếng anh 11 Global Success UNIT 1 - Bản HS.doc
Bài tập - Tiếng anh 11 Global Success UNIT 1 - Bản HS.docBài tập - Tiếng anh 11 Global Success UNIT 1 - Bản HS.doc
Bài tập - Tiếng anh 11 Global Success UNIT 1 - Bản HS.doc
 
The Influence of Marketing Strategy and Market Competition on Business Perfor...
The Influence of Marketing Strategy and Market Competition on Business Perfor...The Influence of Marketing Strategy and Market Competition on Business Perfor...
The Influence of Marketing Strategy and Market Competition on Business Perfor...
 
Set off and carry forward of losses and assessment of individuals.pptx
Set off and carry forward of losses and assessment of individuals.pptxSet off and carry forward of losses and assessment of individuals.pptx
Set off and carry forward of losses and assessment of individuals.pptx
 
Recruiting in the Digital Age: A Social Media Masterclass
Recruiting in the Digital Age: A Social Media MasterclassRecruiting in the Digital Age: A Social Media Masterclass
Recruiting in the Digital Age: A Social Media Masterclass
 
Project File Report BBA 6th semester.pdf
Project File Report BBA 6th semester.pdfProject File Report BBA 6th semester.pdf
Project File Report BBA 6th semester.pdf
 
ikea_woodgreen_petscharity_cat-alogue_digital.pdf
ikea_woodgreen_petscharity_cat-alogue_digital.pdfikea_woodgreen_petscharity_cat-alogue_digital.pdf
ikea_woodgreen_petscharity_cat-alogue_digital.pdf
 
Cree_Rey_BrandIdentityKit.PDF_PersonalBd
Cree_Rey_BrandIdentityKit.PDF_PersonalBdCree_Rey_BrandIdentityKit.PDF_PersonalBd
Cree_Rey_BrandIdentityKit.PDF_PersonalBd
 
Brand Analysis for an artist named Struan
Brand Analysis for an artist named StruanBrand Analysis for an artist named Struan
Brand Analysis for an artist named Struan
 
Mastering B2B Payments Webinar from BlueSnap
Mastering B2B Payments Webinar from BlueSnapMastering B2B Payments Webinar from BlueSnap
Mastering B2B Payments Webinar from BlueSnap
 
Putting the SPARK into Virtual Training.pptx
Putting the SPARK into Virtual Training.pptxPutting the SPARK into Virtual Training.pptx
Putting the SPARK into Virtual Training.pptx
 
Call 8867766396 Satta Matka Dpboss Matka Guessing Satta batta Matka 420 Satta...
Call 8867766396 Satta Matka Dpboss Matka Guessing Satta batta Matka 420 Satta...Call 8867766396 Satta Matka Dpboss Matka Guessing Satta batta Matka 420 Satta...
Call 8867766396 Satta Matka Dpboss Matka Guessing Satta batta Matka 420 Satta...
 
Maksym Vyshnivetskyi: PMO Quality Management (UA)
Maksym Vyshnivetskyi: PMO Quality Management (UA)Maksym Vyshnivetskyi: PMO Quality Management (UA)
Maksym Vyshnivetskyi: PMO Quality Management (UA)
 
Evgen Osmak: Methods of key project parameters estimation: from the shaman-in...
Evgen Osmak: Methods of key project parameters estimation: from the shaman-in...Evgen Osmak: Methods of key project parameters estimation: from the shaman-in...
Evgen Osmak: Methods of key project parameters estimation: from the shaman-in...
 
BeMetals Investor Presentation_June 1, 2024.pdf
BeMetals Investor Presentation_June 1, 2024.pdfBeMetals Investor Presentation_June 1, 2024.pdf
BeMetals Investor Presentation_June 1, 2024.pdf
 
FINAL PRESENTATION.pptx12143241324134134
FINAL PRESENTATION.pptx12143241324134134FINAL PRESENTATION.pptx12143241324134134
FINAL PRESENTATION.pptx12143241324134134
 
In the Adani-Hindenburg case, what is SEBI investigating.pptx
In the Adani-Hindenburg case, what is SEBI investigating.pptxIn the Adani-Hindenburg case, what is SEBI investigating.pptx
In the Adani-Hindenburg case, what is SEBI investigating.pptx
 

Concrete compressive strength

  • 1. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 1/34 Concrete compressive strength We have a numerical dataset with information about di erent batches of concrete. Each batch is Strength of Concrete in speci c moment in time (number of days after prodaction). It is known which and how much components are in of Concrete. There are information about 1030 batches in the dataset. Cement Slag Ash Water Plasticizer Breakstone Sand Time Strength 540.0 0.0 0 162 2.5 1040.0 676.0 28 79.99 540.0 0.0 0 162 2.5 1055.0 676.0 28 61.89 332.5 142.5 0 228 0.0 932.0 594.0 270 40.27 332.5 142.5 0 228 0.0 932.0 594.0 365 41.05 198.6 132.4 0 192 0.0 978.4 825.5 360 44.30 Main components for production: Cement, Water, Sand, Breakstone. These components are always using for Concrete production. Additional components: Not all batches contain Slag, Ash or Plasticizer. Plasticizer is used to add less water. If we add Ash, we can add less Cement. Furnace Slag in uences on the concrete structure. Time: There are bathes with the same composition of components, but with di erent number of days. The Concrete Strength has been checked after 1,3, 7, 14, 28, 56, 90, 91, 100, 120, 180, 270, 360, 365 days after prodaction.There are a few data observations where Time is 1, 120 days in the dataset. We have exclude this observation from consideration. We have little data with Time = 90, 91, 100 and Time = 270, 360, 365. We use 90% of all batches. When, its strength has reached 99% in 28 days, still concrete continues to gain strength after that period, but that rate of gain in compressive strength is very less compared to that in 28 days. So, data with Time = 90, 91 we have set Time = 90 and observations with Time = 270, 360, 365 we have set Time = 365. So, there are 1030 observations in the dataset, we use 1025. 1m3 1m3
  • 2. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 2/34 Our task: determine how a change in the proportions of di erent elements a ects the commpresive strength of concrete. Base analysis Base vizualization We can see strong dependence between Cement and Strength. The positive value of the slope in the Linear Model (Strength~Cement) shows that If we add more Cement, Concrete Strength increases. The negative values of the slope in the Linear Models (Strength~Water, Strength~Sand) show that It is inverse dependence between Strength and Water; Strength and Sand. The value of the slope in the Breakstone~Strength less than other, it shows that Breakstone practically doesn’t a ect on Strength.
  • 3. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 3/34 We can see inverse dependence between Ash and Strength (the negative value of the slope). The value of the slope in the Slag~Strength less than other, it shows that It is low dependence between Slag and Strength. The positive value of the slope in the Plasticizer~Strength model indicates that an increase in quantity Plasticizer improves Strength. Correlation matrix It shows that there isn’t big intercorrelation (>0.7) between explanatory variables and we don’t need to delete any variables. It is easy to see that the most in uential components are Time and Cement.We can see strong intercorrelation between Water and Plasticizer. If we add Plasticizer, we can use less Water. Before building a model we have checked for normality the dependent variable (Strength). This Strength doesn’t Normally Distributed, but for Linear Regression this is an optional condition. Building models Hypothesis 1: Concrete compressive strength depends on all ingredients and Time. ##
  • 4. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 4/34 ## ## Call: ## lm(formula = Strength ~ ., data = data) ## ## Residuals: ## Min 1Q Median 3Q Max ## -28.970 -6.455 0.403 6.875 34.255 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) -25.686533 27.145205 -0.946 0.344239 ## Cement 0.120191 0.008657 13.883 < 2e-16 *** ## Slag 0.103565 0.010326 10.029 < 2e-16 *** ## Ash 0.088255 0.012815 6.887 9.99e-12 *** ## Water -0.146563 0.040955 -3.579 0.000362 *** ## Plasticizer 0.291271 0.094866 3.070 0.002195 ** ## Breakstone 0.019292 0.009598 2.010 0.044692 * ## Sand 0.021507 0.010921 1.969 0.049199 * ## Time 0.101555 0.005176 19.620 < 2e-16 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 10.55 on 1016 degrees of freedom ## Multiple R-squared: 0.6043, Adjusted R-squared: 0.6012 ## F-statistic: 193.9 on 8 and 1016 DF, p-value: < 2.2e-16 Cement, Slag, Ash, Water, Time, Breakstone and Sand strongly a ect Strength. The R-squared is 0.6012, it indicates that 60% of the variation in the Strength is explained by the input variables. Model 1 check. Conditionals of Gauss–Markov theorem. 1)Homoscedasticity of the dispersion of residues. p = 0 < 0.05, it means that residual variance is related to model predicted values.Failed 3) An auto-correlation check ( The Durbin–Watson statistic). p = 0 < 0.05, it means that it is auto-correlation. Failed 4)The normal distribution of residues (The Skewness-Kurtosis’s Test).
  • 5. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 5/34 Residues are distributed normally. Sucess
  • 6. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 6/34 Conclusion: Not all conditions of the Gauss–Markov theorem are satis ed. Our hypothesis is false, so Concrete compressive strength doesn’t depend on all ingredients and Time and we can’t use our model. We have made some changes: built Linear Model (Strength ~log(Time)), using regression coe cients, counted predicted values of Strength and after subtraction predicted Strength from primary Strength, we have got values Strength without Time.
  • 7. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 7/34 Hypothesis 2: Concrete compressive strength depends on all ingredients, besides Time. ## ## Call: ## lm(formula = strengthWithoutTime ~ Slag + Ash + Plasticizer + ## Breakstone + Sand + Cement + Water, data = data) ## ## Residuals: ## Min 1Q Median 3Q Max ## -23.7942 -4.4161 -0.5019 3.9322 29.9414 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) -84.013818 18.711087 -4.490 7.93e-06 *** ## Slag 0.113116 0.007079 15.978 < 2e-16 *** ## Ash 0.092166 0.008777 10.500 < 2e-16 *** ## Plasticizer 0.147213 0.065461 2.249 0.0247 * ## Breakstone 0.029070 0.006616 4.394 1.23e-05 *** ## Sand 0.033932 0.007501 4.524 6.79e-06 *** ## Cement 0.133024 0.005965 22.301 < 2e-16 *** ## Water -0.122236 0.028234 -4.329 1.64e-05 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 7.283 on 1017 degrees of freedom ## Multiple R-squared: 0.7301, Adjusted R-squared: 0.7282 ## F-statistic: 392.9 on 7 and 1017 DF, p-value: < 2.2e-16 Cement, Slag, Ash, Water, Time, Breakstone and Sand strongly a ect Strength.The R-squared is 0.7282, it indicates that 72% of the variation in the Strength is explained by the input variables. Model 2 check. Conditionals of Gauss–Markov theorem
  • 8. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 8/34 1)Homoscedasticity of the dispersion of residues. p = 0 < 0.05, it means that residual variance is related to model predicted values. Failed 2) An auto-correlation check ( The Durbin–Watson statistic). p = 0 < 0.05, it means that it is auto-correlation. Failed 3)The normal distribution of residues (The Skewness-Kurtosis’s Test). Residues don’t have the normal distribution. Failed
  • 9. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 9/34 Conclusion: Our hypothesis is false. Model Strength~Time is wrong, because Conditionals of Gauss–Markov theorem are failed. Let’s divide the dataset into two parts: rst part includes observations with main and additional components; second part includes observations only with main components. We have build two models for each sets.
  • 10. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 10/34 Hypothesis 3: Concrete compressive strength depends on only Main components. ## ## Call: ## lm(formula = strengthWithoutTime ~ Cement + Water + Breakstone + ## Sand, data = data_base) ## ## Residuals: ## Min 1Q Median 3Q Max ## -13.0673 -2.8142 -0.0592 2.0592 19.8428 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) 76.471423 28.553201 2.678 0.00802 ** ## Cement 0.099861 0.009644 10.355 < 2e-16 *** ## Water -0.348389 0.047123 -7.393 3.9e-12 *** ## Breakstone -0.023489 0.008764 -2.680 0.00798 ** ## Sand -0.027102 0.012582 -2.154 0.03244 * ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 4.796 on 199 degrees of freedom ## Multiple R-squared: 0.8323, Adjusted R-squared: 0.8289 ## F-statistic: 246.9 on 4 and 199 DF, p-value: < 2.2e-16 Cement, Slag, Ash, Water, Breakstone and Sand strongly a ect Strength, especially Cement and Water.The R-squared is 0.8289, it indicates that 82% of the variation in the Strength is explained by the input variables. Model 3 check. Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues. p = 0.0004 < 0.05, it means that residual variance is related to model predicted values. Failed 2) An auto-correlation check ( The Durbin–Watson statistic). p = 0.016 < 0.05, it means that it is auto-correlation. Failed 3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
  • 11. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 11/34 Residues don’t have the normal distribution. Failed
  • 12. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 12/34 Conclusion: Our hypothesis is false. We can’t use model 3. Model Strength~Time is wrong, because Conditionals of Gauss–Markov theorem are failed. Hypothesis 4: Concrete compressive strength depends on all components (There are additional components in all observations). ## ## Call: ## lm(formula = strengthWithoutTime ~ Cement + Slag + Ash + Water + ## Plasticizer + Breakstone + Sand, data = data_all) ## ## Residuals: ## Min 1Q Median 3Q Max ## -23.210 -4.804 -0.552 4.169 28.024 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) -1.216e+02 2.181e+01 -5.578 3.32e-08 *** ## Cement 1.433e-01 6.953e-03 20.618 < 2e-16 *** ## Slag 1.130e-01 8.595e-03 13.149 < 2e-16 *** ## Ash 9.070e-02 1.046e-02 8.675 < 2e-16 *** ## Water -7.444e-02 3.202e-02 -2.324 0.0203 * ## Plasticizer 9.944e-02 7.315e-02 1.359 0.1744 ## Breakstone 4.513e-02 7.919e-03 5.699 1.68e-08 *** ## Sand 4.719e-02 8.473e-03 5.570 3.47e-08 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 7.357 on 813 degrees of freedom ## Multiple R-squared: 0.7278, Adjusted R-squared: 0.7255 ## F-statistic: 310.6 on 7 and 813 DF, p-value: < 2.2e-16
  • 13. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 13/34 Cement, Slag, Ash, Water, Breakstone and Sand strongly a ect Strength.The R-squared is 0.7255, it indicates that 72% of the variation in the Strength is explained by the input variables. Model 4 check. Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues. p = 0 < 0.05, it means that residual variance is related to model predicted values.Failed 2) An auto-correlation check ( The Durbin–Watson statistic). p = 0 < 0.05, it means that it is auto-correlation. Failed 3)The normal distribution of residues (The Skewness-Kurtosis’s Test). Residues don’t have the normal distribution. Failed
  • 14. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 14/34 Conclusion: Our hypothesis is false. We can’t use model 4. Model Strength~Time is wrong, because Conditionals of Gauss–Markov theorem are failed. All our models are wrong, so we have to apply a di erent analysis approach. We found out and deleted dublicate data (with the same compositon, Time and Strength) and wrong batches. We have got 988 batches. After that, we have grouped batches with di erent composotion, but the same Time. As a result, we have got groups with: Time = 3, 7, 14, 28. Batches with Time = 180,
  • 15. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 15/34 270, 360, 365 were united in one group with Time = 300;with Time = 90, 91, 100 were united in one group with Time = 95, because we have found approximately the same values of Strength for batches with Time = 180, 270, 360, 365 and for Time = 90, 91, 100. We have created 6 datasets with 125, 117, 62, 417, 86, 122, 59 observations, where Time is respectively 3, 7, 14, 28, 56, 95 (90, 91, 100), 300 (180, 270, 360, 365). Time NumberOfBatches 3 125 7 117 14 62 28 417 56 86 95 122 300 59 At rst, we have checked the normal distribution of Strength for di erent groups. It was successful for group with Time = 56, 95, 300, it means that in these datasets Stregth has the normal distribution. But, we will build model for all datasets, because it’s uncritical for Linear Model that predicted variable doesn’t have the normal distribution. Time = 3 Hypothesis: Concrete compressive strength depends on all components. ## ## Call: ## lm(formula = Strength ~ Cement + Sand + Breakstone + Ash + Slag, ## data = d3) ##
  • 16. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 16/34 ## Residuals: ## Min 1Q Median 3Q Max ## -9.3248 -1.9449 -0.0407 2.2557 8.6397 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) -1.398e+02 1.341e+01 -10.422 < 2e-16 *** ## Cement 1.268e-01 5.281e-03 24.015 < 2e-16 *** ## Sand 6.790e-02 7.167e-03 9.474 3.30e-16 *** ## Breakstone 5.815e-02 7.301e-03 7.964 1.11e-12 *** ## Ash 9.115e-02 7.446e-03 12.241 < 2e-16 *** ## Slag 9.537e-02 7.130e-03 13.377 < 2e-16 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 3.701 on 119 degrees of freedom ## Multiple R-squared: 0.859, Adjusted R-squared: 0.8531 ## F-statistic: 145 on 5 and 119 DF, p-value: < 2.2e-16 Cement, Slag, Ash, Breakstone and Sand strongly a ect Strength.The R-squared is 0.8531, it indicates that 85% of the variation in the Strength is explained by the input variables. Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues. p = 0 < 0.05, it means that residual variance is related to model predicted values.Failed 2) An auto-correlation check ( The Durbin–Watson statistic). p = 0.02,it means that it isn’t auto-correlation. Sucess 3)The normal distribution of residues (The Skewness-Kurtosis’s Test). Residues have the normal distribution. Sucess
  • 17. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 17/34 Conclusion: Our hypothesis is false. Because one contitional of Gauss–Markov theorem is failed. Time = 7 Hypothesis: Concrete compressive strength depends on all components. ## ## Call:
  • 18. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 18/34 ## Call: ## lm(formula = Strength ~ Cement + Water + Plasticizer + Slag - ## 1, data = d7) ## ## Residuals: ## Min 1Q Median 3Q Max ## -15.0102 -2.8381 -0.4443 2.5889 13.4414 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## Cement 0.105573 0.005572 18.947 < 2e-16 *** ## Water -0.076393 0.010230 -7.468 1.84e-11 *** ## Plasticizer 0.590486 0.091078 6.483 2.44e-09 *** ## Slag 0.052515 0.005285 9.936 < 2e-16 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 4.614 on 113 degrees of freedom ## Multiple R-squared: 0.9757, Adjusted R-squared: 0.9749 ## F-statistic: 1137 on 4 and 113 DF, p-value: < 2.2e-16 Cement, Slag, Water, Plasticizer strongly a ect Strength.The R-squared is 0.9749, it indicates that 97.5% of the variation in the Strength is explained by the input variables. Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues. p = 0 < 0.05, it means that residual variance is related to model predicted values.Failed 2) An auto-correlation check ( The Durbin–Watson statistic). p = 0.378,it means that it isn’t auto-correlation. Sucess 3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
  • 19. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 19/34 Residues have the normal distribution. Sucess
  • 20. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 20/34 Conclusion: Our hypothesis is false. Because Homoscedasticity of the dispersion of residues is failed. Time = 14 Hypothesis: Concrete compressive strength depends on all components. ## ## Call: ## lm(formula = Strength ~ Cement + Water + Sand + Plasticizer + ## Ash + Slag, data = d14) ## ## Residuals: ## Min 1Q Median 3Q Max ## -11.9751 -2.4389 -0.3798 2.8660 7.9254 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) -42.39184 20.43416 -2.075 0.04272 * ## Cement 0.16392 0.01685 9.726 1.50e-13 *** ## Water -0.09603 0.04460 -2.153 0.03570 * ## Sand 0.03811 0.01422 2.680 0.00968 ** ## Plasticizer 0.60109 0.20272 2.965 0.00447 ** ## Ash 0.10555 0.02217 4.760 1.45e-05 *** ## Slag 0.14019 0.02108 6.650 1.41e-08 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 4.03 on 55 degrees of freedom ## Multiple R-squared: 0.8038, Adjusted R-squared: 0.7824 ## F-statistic: 37.55 on 6 and 55 DF, p-value: < 2.2e-16
  • 21. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 21/34 Cement, Water, Slag, Ash, Plasticizer, Sand strongly a ect Strength.The R-squared is 0.7824, it indicates that 78.2% of the variation in the Strength is explained by the input variables. Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues. p = 0.051289, it means that residual variance isn’t related to model predicted values.Sucess 2) An auto-correlation check ( The Durbin–Watson statistic). p = 0.742,it means that it isn’t auto-correlation. Sucess 3)The normal distribution of residues (The Skewness-Kurtosis’s Test). Residues have the normal distribution. Sucess
  • 22. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 22/34 Conclusion: Our hypothesis is true. It means that Least squares is better for this data and we can use this model. Time = 28 Hypothesis: Concrete compressive strength depends on all components. ## ## Call:
  • 23. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 23/34 ## Call: ## lm(formula = Strength ~ Cement + Water + Sand + Breakstone + ## Ash + Slag, data = d28) ## ## Residuals: ## Min 1Q Median 3Q Max ## -20.646 -4.046 -1.357 3.436 27.835 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) -79.647940 27.560586 -2.890 0.00406 ** ## Cement 0.165211 0.009482 17.423 < 2e-16 *** ## Water -0.090868 0.035966 -2.526 0.01190 * ## Sand 0.049274 0.011697 4.213 3.11e-05 *** ## Breakstone 0.034402 0.009934 3.463 0.00059 *** ## Ash 0.104380 0.014159 7.372 9.36e-13 *** ## Slag 0.139517 0.011267 12.383 < 2e-16 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 7.104 on 410 degrees of freedom ## Multiple R-squared: 0.7609, Adjusted R-squared: 0.7574 ## F-statistic: 217.4 on 6 and 410 DF, p-value: < 2.2e-16 Cement, Slag, Water, Plasticizer, Sand, Breakstone, Ash strongly a ect Strength.The R-squared is 0.7574, it indicates that 75% of the variation in the Strength is explained by the input variables. Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues. p = 0 < 0.05, it means that residual variance is related to model predicted values.Failed 2) An auto-correlation check ( The Durbin–Watson statistic). p = 0,it means that it is auto-correlation. Failed 3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
  • 24. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 24/34 Residues don’t have the normal distribution. Failed
  • 25. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 25/34 Conclusion: Our hypothesis is false. All condidtionals of Gauss–Markov theorem are failed. We have to split into several parts this dataset and build the model again. Time = 56 Hypothesis: Concrete compressive strength depends on all components. ## ## Call: ## lm(formula = Strength ~ Cement + Water + Sand + Ash + Slag - ## 1, data = d56) ## ## Residuals: ## Min 1Q Median 3Q Max ## -18.7469 -4.0671 -0.8282 4.7413 16.7102 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## Cement 0.103363 0.008088 12.779 < 2e-16 *** ## Water -0.140737 0.029632 -4.749 8.69e-06 *** ## Sand 0.032485 0.005855 5.548 3.53e-07 *** ## Ash 0.117464 0.017629 6.663 2.99e-09 *** ## Slag 0.161311 0.012577 12.826 < 2e-16 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 6.468 on 81 degrees of freedom ## Multiple R-squared: 0.9857, Adjusted R-squared: 0.9848 ## F-statistic: 1118 on 5 and 81 DF, p-value: < 2.2e-16 Cement, Water, Slag, Ash, Sand strongly a ect Strength.The R-squared is 0.9848, it indicates that 98% of the variation in the Strength is explained by the input variables. Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues.
  • 26. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 26/34 p = 0.57373, it means that residual variance isn’t related to model predicted values.Sucess 2) An auto-correlation check ( The Durbin–Watson statistic). p = 0.11,it means that it isn’t auto-correlation. Sucess 3)The normal distribution of residues (The Skewness-Kurtosis’s Test). Residues have the normal distribution. Sucess
  • 27. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 27/34 Conclusion: Our hypothesis is true. It means that Least squares is better for this data and we can use this model. Time = 95 Hypothesis: Concrete compressive strength depends on all components. ## ## Call:
  • 28. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 28/34 ## Call: ## lm(formula = Strength ~ Cement + Water + Ash + Slag, data = d95) ## ## Residuals: ## Min 1Q Median 3Q Max ## -18.8475 -3.0299 -0.7854 3.3528 16.1291 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) 40.938765 7.004618 5.845 4.68e-08 *** ## Cement 0.110679 0.007575 14.611 < 2e-16 *** ## Water -0.202701 0.028434 -7.129 9.00e-11 *** ## Ash 0.113753 0.015122 7.522 1.20e-11 *** ## Slag 0.109286 0.008733 12.514 < 2e-16 *** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 6.071 on 117 degrees of freedom ## Multiple R-squared: 0.7829, Adjusted R-squared: 0.7755 ## F-statistic: 105.5 on 4 and 117 DF, p-value: < 2.2e-16 Cement, Water, Slag, Ash strongly a ect Strength.The R-squared is 0.7755, it indicates that 77% of the variation in the Strength is explained by the input variables. Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues. p = 0.067624, it means that residual variance isn’t related to model predicted values.Sucess 2) An auto-correlation check ( The Durbin–Watson statistic). p = 0.476,it means that it isn’t auto-correlation. Sucess 3)The normal distribution of residues (The Skewness-Kurtosis’s Test).
  • 29. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 29/34 Residues have the normal distribution. Sucess
  • 30. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 30/34 Conclusion: Our hypothesis is true. It means that Least squares is better for this data and we can use this model. Time = 300 Hypothesis: Concrete compressive strength depends on all components. ## ## Call: ## lm(formula = Strength ~ Cement + Sand + Breakstone + Slag, data = d300) ## ## Residuals: ## Min 1Q Median 3Q Max ## -14.974 -4.686 -1.429 6.205 14.222 ## ## Coefficients: ## Estimate Std. Error t value Pr(>|t|) ## (Intercept) 89.06778 31.56549 2.822 0.00667 ** ## Cement -0.15165 0.05226 -2.902 0.00536 ** ## Sand -0.17691 0.04058 -4.360 5.89e-05 *** ## Breakstone 0.14330 0.02781 5.152 3.73e-06 *** ## Slag -0.15096 0.05292 -2.852 0.00614 ** ## --- ## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 ## ## Residual standard error: 7.113 on 54 degrees of freedom ## Multiple R-squared: 0.5813, Adjusted R-squared: 0.5503 ## F-statistic: 18.74 on 4 and 54 DF, p-value: 1.034e-09 Cement, Water, Slag, Ash strongly a ect Strength.The R-squared is 0.5503 , it indicates that 55% of the variation in the Strength is explained by the input variables. Conditionals of Gauss–Markov theorem 1)Homoscedasticity of the dispersion of residues.
  • 31. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 31/34 p = 0.13509, it means that residual variance isn’t related to model predicted values.Sucess 2) An auto-correlation check ( The Durbin–Watson statistic). p = 0,it means that it is auto-correlation. Failed 3)The normal distribution of residues (The Skewness-Kurtosis’s Test). Residues have the normal distribution. Sucess
  • 32. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 32/34 Conclusion: Our hypothesis is false. Because An auto-correlation check is failed.
  • 33. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 33/34 If we use model for Time = 14 for all data, we get such result.
  • 34. 17.08.2020 Concrete compressive strength 212.3.101.119:3890 34/34