Machine Learning Project 2 - Adult Income Data

By


This blog post contains the 2nd part of this project, i.e. Model Building.
Here 2 models will be built: Linear Model (i.e. Logistic Regression) and Non-Linear Model (i.e. Random Forest)
Follow the 1st part, i.e Data Analysis  --> Click Here

 Model Building:
1. Linear Model

 Logistic Regression:
If we look at the overall F1-score, this linear model managed to achieve 0.85, which is good.

But, the gap between the recall values is a lot. LogisticRegression managed to perform far better in predicting class type 0 (i.e. <=50K) with 0.93, but for class type 1 (i.e. >50K) it is very poor (only 0.59).

 Lets look at the coefficients:
The top 3 features that contributes positively are:
1. Capital Gain
2. Education
3. Age

 The top 3 features that contributes negatively are:
1. Marital Status - Others
2. Occupation - Others
3. Sex - Female

Quick recap from the Data Analysis post:
Positive contributions:
1. Capital Gain - If capital gains present, then higher chances of getting >50K.
2. Education    -  higher the education, higher the chances of getting >50K.
3. Age              - higher the age, higher the chances of getting >50K.

 Negative Contributions:
1. Marital Status - Others : This group has no chance/very little chance of getting >50K.
2. Occupation - Others    :  no chance/very little chance of getting >50K.
3. Sex - Female               : low chance of getting >50K as compared to Males. Hence Sex-Male has a positive coefficient.

2. Non-Linear Model
Random Forest:
This non-linear model managed to achieve F1-score of 0.96, which is fantastic.
From the recall values, it is not only predicting the class type 0 nicely with 0.98, also managed to predict class type 1 (i.e. 0.89) far better than the linear model.
Lets look at the Feature Importances:
Features with the most Importance score:
1. Age
2. Education
3. Capital Gain
4. Hours per week

 NOTE: As we saw from the data analysis part, higher the numbers in these features, higher the chances of getting >50K.

Comparing Classifiers:

 Conclusion:
The non-linear model (i.e. RandomForest) performs far better than the linear model (i.e. LogisticRegression) when it comes to predicting both the classes of the Target variable, as can be seen from the recall-values.
From the ROC curves, we get a nearly perfect curve for RandomForest.
To view the full code on Github --> Click Here

Comments

Post a Comment

Popular posts from this blog

A/B Testing: Bernoulli Model VS the Beta-Binomial Hierarchical Model

By
Image
What is A/B Testing? A/B testing  is a method of comparing two versions of a webpage or app against each other to determine which one performs better.  AB testing is essentially an experiment where two or more variants of a page are shown to users at random, and based on the data it receives, statistical analysis is used to determine which variation performs better for a given conversion goal. How A/B Testing works? Here, 2 variations of the same webpage or app are created and are randomly assigned to the users (say 50% to Variation A and 50% to Variation B).  As users are served either the A or B, their engagement with each variation is measured and collected and analyzed through a statistical engine. You can then determine whether changing the webpage or app experience had a positive, negative, or no effect on user behavior. Why A/B Testing should be used? A/B testing allows individuals, teams, and companies to make careful changes to their user experie
Continue Reading »

Exploratory Data Analysis and Hypothesis Testing - Loan Prediction

By
Image
Problem Statement: About Company Dream Housing Finance company deals in all home loans. They have presence across all urban, semi urban and rural areas. Customer first apply for home loan after that company validates the customer eligibility for loan. Problem Company wants to automate the loan eligibility process (real time) based on customer detail provided while filling online application form. These details are Gender, Marital Status, Education, Number of Dependents, Income, Loan Amount, Credit History and others. To automate this process, they have given a problem to identify the customers segments, those are eligible for loan amount so that they can specifically target these customers.  Introduction The objective of this project is to do “Exploratory Analysis” and “Hypothesis Testing” on the features of this dataset to find various insights as to how each feature affects the chances of getting a loan. Dataset The Loan Prediction dataset consists of 613 cust
Continue Reading »

Recurrent Neural Networks and LSTM explained

By
Image
What are Recurrent Neural Networks ? Recurrent Neural Networks are the state of the art algorithm for sequential data  This is because it is the first algorithm that remembers its input, due to an internal memory, which makes it perfectly suited for Machine Learning problems that involve sequential data.   In simple terms, they are networks with loops in them, allowing information to be saved. Here, A is the network, Xt is the input with output  ht.  Although, these loops makes the RNN kind of hard to interpret, but in reality it is very simple. This is how the RNN looks when we unroll them. A RNN can be thought of as multiple copies of the same network , each passing message to  the next. Because of their internal memory, RNN’s are able to remember important things about the input they received, which enables them to be very precise in predicting what’s coming next. This is the reason why they are the preferred algorithm for sequential data like time
Continue Reading »