Project 1

OBJECTIVE: Logistic regression to be performed with simple random sampling & stratified sampling and identify the differences between the two sampling methods.

SOLUTION PROCEDURE:

Logistics regression is to make the predict the target value with the inputs of the feature matrix, they use the line equation for the prediction.

• Importing the required libraries & functions
• Loading the data into the data frame & taking one of the faults as target (y)& features as (x).
• Feature & target matrix are split into test & train data to get better understanding of the trained model with the simple random sampling & stratified sampling.
• StandardScaler () function is used to scale the model since the logistic regression is affected by the scale of the data inputs
• With the help of inbuilt functions, parameters are obtained for the training data set & then with those data, predictions are made for test data.
• We determine the accuracy & time taken for the simple random sampling & stratified sampling.
• We also plotting the test data with the predicted data with simple random & stratified random sampling.
• We plot the confusion matrix to understand in detail how the prediction of target data happened with simple random sampling & stratified sampling.

CODE & OUTPUT: Jupyter notebook attached

CONCLUSION:

Based on comparison we have found the pros & cons with simple and random sampling.

Simple random sampling:

Pros: Slightly higher accuracy prediction which is shown by the score & the number of false negatives in the confusion matrix

Cons:7 samples reported as False positive since a fault is reported as not a fault

Stratified sampling:

Pros: Significant reduction in the time taken to fit the model

Cons: 48 samples are reported as False positive which is over protective compared to the simple random sampling but not bad as compared to having results of False positive