The document discusses preparing data for machine learning by transforming raw data into machine learning-ready data. It outlines a holistic approach that involves defining goals, understanding required data structures, assessing available data, and performing transformations like cleaning, denormalizing, aggregating, pivoting, and feature engineering. The transformations are aimed at structuring the data into a format that machine learning algorithms can consume to build models. Automating the transformations and evaluating results is also emphasized.

1. September 8-9, 2016

2. BigML, Inc 2
Basic Transformations
Poul Pertesen
CIO, BigML, Inc
Creating Machine Learning Ready Data

3. BigML, Inc 3Machine Learning-Ready Data
Basic Transformations
Q: How does a physicist milk a cow?
A: Well, first let us consider a spherical cow...
Q: How does a data scientist build a model?
A: Well, first let us consider perfectly formatted data…

4. BigML, Inc 4Machine Learning-Ready Data
The Dream
CSV Dataset Model Profit!

5. BigML, Inc 5Machine Learning-Ready Data
The Reality
CRM
Web Accounts
Transactions
ML Ready?
Is all hope lost?
How do you even start?

6. BigML, Inc 6Machine Learning-Ready Data
Holistic Approach
• Define a clear idea of the goal.
• Understand what ML tasks will achieve the goal.
• Understand the data structure to perform those ML tasks.
• Find out what kind of data you have and make it ML-Ready
• where is it, how is it stored?
• what are the features?
• can you access it programmatically?
• Feature Engineering: transform the data you have into the
data you actually need.
• Evaluate: Try it on a small scale
• Accept that you might have to start over….
• But when it works, automate it!!!!

7. BigML, Inc 7Machine Learning-Ready Data
Holistic Approach
Define Goal & ML Task

8. BigML, Inc 8Machine Learning-Ready Data
Understand ML Tasks
Goal
• Will this customer default on a loan?
• How many customers will apply for a
loan next month?
• Is the consumption of this product
unusual?
• Is the behavior of the customers
similar?
• Are these product purchased
together?
ML Task
Classification
Regression
Anomaly Detection
Cluster Analysis
Association Discovery

9. BigML, Inc 9Machine Learning-Ready Data
Holistic Approach
Required Data Structure

10. BigML, Inc 10Machine Learning-Ready Data
Classification
CategoricalTrainingTesting
Predicting

11. BigML, Inc 11Machine Learning-Ready Data
Regression
NumericTrainingTesting
Predicting

12. BigML, Inc 12Machine Learning-Ready Data
Anomaly Detection

13. BigML, Inc 13Machine Learning-Ready Data
Cluster Analysis

14. BigML, Inc 14Machine Learning-Ready Data
Association Discovery

15. BigML, Inc 15Machine Learning-Ready Data
Holistic Approach
Make Your Data ML-Ready

16. BigML, Inc 16Machine Learning-Ready Data
ML-Ready Data
Instances
Fields
(Features)
Tabular Data:
• Each row is one of the instances.
• Each column is a field that describes a property of the
instance that is relevant to the question being modeled.
• Fields can be:
already be present in your data
derived from your data
or generated using other fields.
Machine Learning
Algorithms consume
instances of the
question that you want
to model.
!! Danger Ahead !!

17. BigML, Inc 17Machine Learning-Ready Data
Cleansing
Homogenize missing values and different types in the same
feature, fix input errors, correct semantic issues, types, etc.
Name Date Duration (s) Genre Plays
Highway star 1984-05-24 - Rock 139
Blues alive 1990/03/01 281 Blues 239
Lonely planet 2002-11-19 5:32s Techno 42
Dance, dance 02/23/1983 312 Disco N/A
The wall 1943-01-20 218 Reagge 83
Offside down 1965-02-19 4 minutes Techno 895
The alchemist 2001-11-21 418 Bluesss 178
Bring me down 18-10-98 328 Classic 21
The scarecrow 1994-10-12 269 Rock 734
Original
data
Name Date Duration (s) Genre Plays
Highway star 1984-05-24 Rock 139
Blues alive 1990-03-01 281 Blues 239
Lonely planet 2002-11-19 332 Techno 42
Dance, dance 1983-02-23 312 Disco
The wall 1943-01-20 218 Reagge 83
Offside down 1965-02-19 240 Techno 895
The alchemist 2001-11-21 418 Blues 178
Bring me down 1998-10-18 328 Classic 21
The scarecrow 1994-10-12 269 Rock 734
Cleaned
data

18. BigML, Inc 18Machine Learning-Ready Data
Denormalizing
users
artists
tracks
albums
Instances
Features
(millions)
join
Data is usually normalized in relational databases, ML-Ready datasets
need the information de-normalized in a single file/dataset.

19. BigML, Inc 19Machine Learning-Ready Data
Aggregating
User Num.Playbacks Total Time Pref.Device
User001 3 830 Tablet
User002 1 218 Smartphone
User003 3 1019 TV
User005 2 521 Tablet
Aggregated data (list of users)
When the entity to model is different from the provided data, an
aggregation to get the entity might be needed.
Content Genr
e
Duration Play Time User Device
Highway
star
Rock 190 2015-05-12
16:29:33
User001 TV
Blues alive Blues 281 2015-05-13
12:31:21
User005 Tablet
Lonely
planet
Tech
no
332 2015-05-13
14:26:04
User003 TV
Dance,
dance
Disco 312 2015-05-13
18:12:45
User001 Tablet
The wall Reag
ge
218 2015-05-14
09:02:55
User002 Smartphone
Offside
down
Tech
no
240 2015-05-14
11:26:32
User005 Tablet
The
alchemist
Blues 418 2015-05-14
21:44:15
User003 TV
Bring me
down
Class
ic
328 2015-05-15
06:59:56
User001 Tablet
The
scarecrow
Rock 269 2015-05-15
12:37:05
User003 Smartphone
Original data (list of playbacks)
tail -n+2 playlists.csv | cut -d',' -f5 | sort | uniq -c

20. BigML, Inc 20Machine Learning-Ready Data
Pivoting
Different values of a feature are pivoted to new columns in the
result dataset.
Content Genre Duration Play Time User Device
Highway star Rock 190 2015-05-12 16:29:33 User001 TV
Blues alive Blues 281 2015-05-13 12:31:21 User005 Tablet
Lonely planet Techno 332 2015-05-13 14:26:04 User003 TV
Dance, dance Disco 312 2015-05-13 18:12:45 User001 Tablet
The wall Reagge 218 2015-05-14 09:02:55 User002 Smartphone
Offside down Techno 240 2015-05-14 11:26:32 User005 Tablet
The alchemist Blues 418 2015-05-14 21:44:15 User003 TV
Bring me down Classic 328 2015-05-15 06:59:56 User001 Tablet
The scarecrow Rock 269 2015-05-15 12:37:05 User003 Smartphone
Original data
User Num.Playback
s
Total Time Pref.Device NP_TV NP_Tablet NP_Smartphone TT_TV TT_Tablet TT_Smartphone
User001 3 830 Tablet 1 2 0 190 640 0
User002 1 218 Smartphone 0 0 1 0 0 218
User003 3 1019 TV 2 0 1 750 0 269
User005 2 521 Tablet 0 2 0 0 521 0
Aggregated data with pivoted columns

21. BigML, Inc 21Machine Learning-Ready Data
Time Windows
Create new features using values over different periods of time
Instances
Features
Time
Instances
Features
(millions)
(thousands)
t=1 t=2 t=3

22. BigML, Inc 22Machine Learning-Ready Data
Updates
Need a current view of the data, but new data only comes in
batches of changes
day
1day
2day
3
Instances
Features

23. BigML, Inc 23Machine Learning-Ready Data
Structuring Output
• A CSV file uses plain text to store tabular data.
• In a CSV file, each row of the file is an instance.
• Each column in a row is usually separated by a comma (,) but other
"separators" like semi-colon (;), colon (:), pipe (|), can also be used. Each
row must contain the same number of fields
• but they can be null
• Fields can be quoted using double quotes (").
• Fields that contain commas or line separators must be quoted.
• Quotes (") in fields must be doubled ("").
• The character encoding must be UTF-8
• Optionally, a CSV file can use the first line as a header to provide the
names of each field.
After all the data transformations, a CSV (“Comma-Separated
Values) file has to be generated, following the rules below:

24. BigML, Inc 24Machine Learning-Ready Data
Holistic Approach
Feature Engineering

25. BigML, Inc 25Machine Learning-Ready Data
Feature Engineering
• Flatline
• Domain Specific Language for data generation
and filtering
• Works with datasets -> datasets
• Lots of built-in functions
• Sliding windows
• Date/Time parsing
• Flatline Editor (in UI)
• https://github.com/bigmlcom/flatline

26. BigML, Inc 26Machine Learning-Ready Data
Feature Engineering
• Feature Engineering of Numeric features:
• Discretization (percentiles, within percentiles, groups)
• Replacement
• Normalization
• Exponentiation, Logarithms, Squares, etc.
• Shock
• Feature Engineering of Text features:
• Misspellings
• Length
• Number of subordinate sentences
• Language
• Levenshtein distance
• Stacking:
• Compute a field using non-linear combinations of other fields

27. BigML, Inc 27Machine Learning-Ready Data
Holistic Approach
Test & Automate

28. BigML, Inc 28Machine Learning-Ready Data
Test & Automate
• Test - Evaluate
• Did you meet the goal?
• If not, did you discover something else useful?
• If not, start over
• If you did…
• Automate - You don’t want to hand code that every time, right?
• Consider tools that are easy to automate
• scripting interface
• APIs
• Ability to maintenance is important

29. BigML, Inc 29Machine Learning-Ready Data
Tools
• Command Line?
• join, cut, awk, sed, sort, uniq
• Automation
• Shell, Python, etc
• Talend
• BigML: bindings, bigmler, API, whizzml
• Relational DB
• MySQL
• Non-Relational DB
• MongoDB

30. BigML, Inc 30Machine Learning-Ready Data
Prosper
Submit Bids
Cancelled Withdraw
Funded
Expired
Defaulted
Paid
Current
Late
Q: Which new loans make it to funded?
Q: Which funded loans make it to paid?
Q: If funded, what will be the rate?
Classification
Regression
Classification

31. BigML, Inc 31Machine Learning-Ready Data
Prosper
Data Provided in XML updates!!
fetch.sh
“curl”
daily
export.sh
import.py
XML
bigml.sh
Model
Predict
Share in gallery
Status
LoanStatus
BorrowerRate

32. BigML, Inc 32Machine Learning-Ready Data
Prosper
• XML… yuck!
• MongoDB has CSV export and is record based so it is easy to
handle changing data structure.
• Feature Engineering
• There are 5 different classes of “bad” loans
• Date cleanup
• Type casting: floats and ints
• Would be better to track over time
• number of late payments
• compare predictions and actuals
• XML… yuck!
Tidbits and Lessons Learned….

33. BigML, Inc 33Machine Learning-Ready Data
Diabetes
Fix Missing Values in a “Meaningful” Way
Filter Zeros
Model
insulin
Predict
insulin
Select
insulin
Fixed
Dataset
Amended
Dataset
Original
Dataset
Clean
Dataset

34. BigML, Inc 34Machine Learning-Ready Data
Stock Prices
(/ (- ( f "price") (avg-window "price" -4, -1)) (standard-deviation "price"))
Shock: Deviations from Trend
date volume price
1 34353 314
2 44455 315
3 22333 315
4 52322 321
5 28000 320
6 31254 319
7 56544 323
8 44331 324
9 81111 287
10 65422 294
11 59999 300
12 45556 302
13 19899 301
14 21453 302
314
314 315
314 315 315
314 315 315 321
315 315 321 320
315 321 320 319
4-Day moving avg)
Current - (4-day avg)
std dev

35. BigML, Inc 35Machine Learning-Ready Data
Talend
https://blog.bigml.com/2013/10/30/data-preparation-for-machine-learning-using-mysql/
Denormalization Example

36. BigML, Inc 36Machine Learning-Ready Data
Talend
https://blog.bigml.com/2013/10/30/data-preparation-for-machine-learning-using-mysql/
Denormalization Example