Economic Analysis of USA and India using Arima Model and different classification algorithm to predict recession

1. FINAL PROJECT
PRESENTED BY:
HINA GANDHI
NIKITA ANAND
SHIVAM GOEL

2. OVERVIEW OF PROJECT
• FORECASTING ECONOMIC INDICATORS AND DETERMINING RECESSION
• VISUALIZING THE INDICATORS USING D3 JS AND TABLEAU
• COMPARING ECONOMY OF TWO COUNTRIES USA AND INDIA
• SENTIMENT ANALYSIS ON DEMONETISATION AND RECESSION 2008

3. STEPS
• DATA SCRAPING AND WRANGLING
• FILLING MISSING VALUES
• PARAMETER SELECTION
• BUILDING MODEL(TIME SERIES AND CLASSIFICATION)
• DEPLOY WEB SERVICE
• DEPLOY APPLICATION ON CLOUD

4. ADDING COLUMN RECESSION
• SCRAPED THE DATA FROM LOUIS FRED WEBSITE
• CREATED A NEW COLUMN RECESSION BASED ON % GDP CHANGE.

5. BORUTA
• BORUTA” PACKAGE. BORUTA IS A FEATURE SELECTION ALGORITHM. PRECISELY, IT WORKS AS A
WRAPPER ALGORITHM AROUND RANDOM FOREST.

6. ARIMA MODEL
• STATIONARIZE THE SERIES
• PLOT THE ACF/PACF CHARTS AND FIND OPTIMAL PARAMETERS
• BUILD ARIMA MODEL
• MAKE PREDICTIONS

7. CLASSIFICATION
• THE FORECASTED VALUES FROM ARIMA MODEL ARE USED AS INPUT PARAMETER
TO PREDICT RECESSION IN PARTICULAR QUARTER OF THE YEAR
• RANDOM FOREST
• NEURAL NETWORK
• LOGISTIC REGRESSION

8. SENTIMENT ANALYSIS
SENTIMENTR
THE SENTIMENTR PACKAGE CONTAINS THE SENTIMENT SCORING FUNCTION
‘SENTIMENT’ FROM THIS PACKAGE PRODUCES RESULTS WITH A GREAT QUALITY
WITHOUT SACRIFICING THE PERFORMANCE BASED ON OUR OBSERVATION . IT
GOES BEYOND A SIMPLE ‘WORD-TO-SENTIMENT’ DICTIONARY APPROACH AND
TAKES INTO ACCOUNT CONTEXTUAL VALENCE SHIFTERS, SUCH
AS NEGATIONS AND INTENSIFIERS.

9. RECESSION 2008

10. DEMONETIZATION

11. THANK YOU