Machine Learning Exercise

Question 1:

The dataset contains the collection that appeared on the Reuters newswire in 1987. The documents were assembled and indexed with categories by personnel from Reuters Ltd.

Formatting of the documents and production of associated data files was done in 1990 by Devid D. Lewis and Stephen Harding at the Information Retrieval Laboratory. The articles are related to category Money and crude which are target classes

Problem statements:

• Load the dataset from the location of the file provided in “input Format” using pandas

• Convert the article_text into word tokens for word classes (money, crude) individually

• Find total number of unique tokens for money

• Find total number of unique tokens of crude

• Convert the article_text into word tokens for both target classes

• Remove stop words and punctuations(use reuters_stopwords provided in stub)

• Lemmatize and compute the size of the vocabulary

• Transform article_text into TF-IDF vector using following parameters:

• Find the size of TF-IDF vector and save as (x,y)

• Save the findings in a file named output.csv at the location/code/output/output.csv

Hint: While converting articles into tokens join them using single whitespace

Input format:

Read data from a file named Reuters.csv at the location/data/training/Reuters.csv

Output Format:

The output file output.csv should contain the findings in the following order:

• The 1st row contains the total number of unique tokens for money.

• The 2nd row contains the total number of unique tokens for crude

• The 3rd row contains the size of vocabulary after lemmatization

• The 4rth row contains the size of TF-IDF vector in a tuple format i.e. (x,y)

Question 2:

Description

The dataset consists of details of various applications along with their eligibility for loan. The details are:

Problem statements:

Based on this dataset, write a python code to perform the following operations:

• Load the dataset from the location of the file provided in “input format” using pandas

• Find the columns with missing data

• Find the number of missing values in each column with missing data

• Create the pandas' data frame with the columns: Column, Missing_value, Impute_method, and fill the data as:

• Sava the columns in ascending order of missing values and imputing method in a file named Impute.csv the location code/output/Impute.csv

• Find the number of rows present in dataset

• Find the number of duplicate rows in dataset

• Calculate the default correlation between the numeric columns

• Find the highest correlation value in dataset (up to 3 decimal places)

• Remove all rows with missing data

• Split the dataset into Train-Test with the parameter test-size = 0.2 and random_state = 0

• Find number of rows in test set

• Save the findings in file named output.csv at the location code/output/output.csv

Input format:

Read data from filename BankLoan.csv present at the location /data/training/BankLaon.csv

Output Format:

The output file output.csv should contain the findings in the following order:

• The first rows contain the number of rows present in the dataset

• The second row contains the number of duplicate rows in the dataset

• The third row contains the highest correlation value in the dataset(upto 3 decimal places)

• The fourth row contains the number of rows in the test set

Question 3:

The dataset has the details of 112 matches along with the date. The matches were between 8 countries and will be held in the country of the home team. The outcome/target variable depicts us either the home team or away wins the match. The features in the dataset are as follows:

Problem statements:

Based on this dataset, write a python code to perform the following operations:

• Load the dataset from the location of the file provided in “input format” using pandas

• Find the mean score in 20 overs of the Home team(Team 1 20over score) up to 3 decimal places

• Find how many matches were won by Home team

• Find the average score of the first batsman (Team 1 Batsman 1 score) when India is playing as Home Team upto 3 decimal places

• Find the average score of the first batsman (Team 1 Batsman 1 Score) when india is batting first up to 3 decimal places

• Drop the date of match column

• Split the target variable(final result) and data(over features)

• Convert Home team, away team, Team who was toss, Team batting first and Team batting first into dummy variable

• Find the number of columns after the conversion

• Using LabelEncoder convert the target variable(Final Result) into binary variable

• Split the dataset into Test-Train with the parameter, test-size = 0.3 and random_state = 0

• Apply logistic regression over the training set

• Find the accuracy of the model

• Save the findings in a file name output.csv at the location/code/output/output.csv

Input format:

Read data from a file named CricketMatchDataset.csv present at the location location/data/training/CricketMatchDataset.csv

Output format:

• The output file output.csv should have the finding in the following order:

• The 1st row contains the mean score in 20 overs of Home Team(Team 1 20ovr Score) upto 3 decimal places

• The 2nd row contains the number of matches were won by Home Team

• The 3rd row contains the average score of the first batsman (Team 1 Batsman 1 score) when india is playing as Home Team up to 3 decimal places

• The 4th row contains the average score of the first batsman (Team 1 Batsman 1 Score) when India is playing as Home Team up to 3 decimal places

• The 5th row contains the number of columns after the conversion

• The 6th row contain the accuracy of the model

Comment below if you need any help to do this practice exercise or if you have a solution then send it on the comment section.