Scribd
Upload
3K views

Logistic Regression Assignment

This document describes a logistic regression assignment to predict whether an individual's annual income is over $50,000 using census data. The data contains over 32,000 records with 15 variables like age, workclass, education, hours worked, native country, and annual income. Questions assess understanding of logistic regression concepts like handling outliers and multicollinearity, and evaluating a logistic regression model in terms of accuracy, specificity, precision, F1 score, and number of correctly classified records.

Uploaded by

Nishant Randev
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
3K views

Logistic Regression Assignment

This document describes a logistic regression assignment to predict whether an individual's annual income is over $50,000 using census data. The data contains over 32,000 records with 15 variables like age, workclass, education, hours worked, native country, and annual income. Questions assess understanding of logistic regression concepts like handling outliers and multicollinearity, and evaluating a logistic regression model in terms of accuracy, specificity, precision, F1 score, and number of correctly classified records.

Uploaded by

Nishant Randev
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Logistic Regression Assignment

[email protected]

+91-7022374614

US: 1-800-216-8930 (Toll-Free)


Problem Statement:

Census-income data plays the most important role in the democratic system of
government, highly affecting the economic sectors. Census-related figures are used
to allocate federal funding by the government to different states and localities.
Census data is also used for post census residents estimates and predictions,
economic and social science research, and many other such applications.
Therefore, the importance of this data and its accurate predictions is very clear to us.
The main aim is to increase awareness about how the income factor actually has an
impact not only on the individual lives of citizens but also an effect on the nation and
its betterment. You will have a look at the data pulled out from the 1994 Census
bureau database, and try to find insights into how various features have an effect on
the income of an individual.
The data contains approximately 32,000 observations with over 15 variables.
The strategy is to analyze the data and perform a predictive task of classification to
predict whether an individual makes over 50K a year or less by using a logistic
regression algorithm.

Column Names Description

Age Age of the individual

Workclass department of the working individual

fnlwgt Final weight of the individual

education The education degree of the individual

education-num Number of years of education

marital-status Marital status of the individual

occupation Occupation of the individual

relationship Relation value

race Ethnicity of the individual

sex Female, Male

capital-gain capital gain of the individual

capital-loss capital loss of the individual


hours-per-week number of working hours

native-country The native country of the individual

Annual-Income Annual income either >50K or <=50K

1. How many types of occupations do we have?


a. 13
b. 14
c. 15
d. 11
2. How many people are working as tech support and have an annual income
greater than 50k?
a. 278
b. 389
c. 289
d. 934
3. How many total missing values are present in the dataset?
a. 4262
b. 5000
c. 5349
d. 4302
4. If there are missing values in the Marital Status column, which option among
the following should be used for replacing the missing values:
a. Mean
b. Median
c. Mode
d. All of the above
5. How many people are having private work classes and are not from the
United States of America?
a. 2151
b. 2300
c. 2000
d. 2190
6. How many people are either having Annual Income(last column) less than or
equal to 50k or their working hours is greater than or equal to 40 hrs:
a. 23008
b. 23448
c. 29505
d. 25903
7. Which of the following methods can you use for handling outliers?
a. Interquartile Range(IQR) Method
b. Z Score method
c. Both of the above methods
d. None of the above
8. Chi-square is used to analyze:
a. Determine the relationship b/w the variables
b. Compare observed results with expected results
c. both a and b
d. None of the above

9. What is VIF?

a. It can detect multicollinearity


b. If the VIF value is greater than 10, then there is no correlation between
the independent variables
c. It stands for Variance Impact Factor
d. VIF is when there is no correlation between one predictor and the other
predictors in a model.

10.What predict_proba will tell you?


a. It will predict the class probabilities
b. It will tell you the target value
c. Both are correct
d. None of the above
11.Logistic regression is useful for regression problems:
a. True
b. False
12.In logistic regression, if the predicted logit is 0, what’s the
transformed probability?
a. 0.5
b. 0.05
c. Both of the above
d. None of the above
13.Which variant of logistic regression is recommended when you have
a categorical dependent variable with more than two values?
a. Multiple Logistic regression
b. Multinomial logistic regression
c. Ordered logit regression
d. Poisson regression

Perform the following tasks for answering the remaining questions

● Rename the last column as Annual Income


● Remove the missing values from the dataset
● Change the labels of categorical data into numerical data using Label
Encoder.
● Split the dataset into a train and test of proportions 70:30 and set the random
state to 0.
● Build a Logistic Regression Model on the data.
Answer the following questions with the help of the above-created model.

14.What is the accuracy score of the above model?


a. 0.60 to 0.70
b. 0.40 to 0.60
c. 0.70 to 0.85
d. None of the above
15.What is the specificity of the above model?
a. 0.20 to 0.30
b. 0.30 to 0.40
c. 0.50 to 0.60
d. None of the above
16.What is the model’s precision when the target is False?
a. 0.60 to 0.70
b. 0.40 to 0.60
c. 0.70 to 0.80
d. None of the above
17.What is the total support value from the above model?
a. 9049
b. 9032
c. 10000
d. 9847
18.What is the f1 score of the above model when the target is
True?
a. 0.30 to 0.40
b. 0.40 to 0.50
c. 0.60 to 0.70
d. 0.90 to 0.99
19.How many records are correctly classified by the
model?
a. 7173
b. 7043
c. 7000
d. None of the above

You might also like