New Assignment 32025
Part 1 Select an appropriate model to train the dataset and make predictions (3 Points)
The UCI Adult dataset-sometimes called the Census Income dataset-is a classic resource in machine learning for demonstrating classification tasks, particularly binary classification.
Dataset Description
·Number of Instances:Around 48,842 rows(depending on whether duplicates/missing rows are handled).
·Number of Attributes:14 features(plus the target)
·Feature Types:
■ Numeric(e.g.,age,hours-per-week,capital-gain).
■ Categorical (e.g.,workclass,marital-status,occupation,sex).
·Target Column:
■ Labeled as income,with possible values >50K or<=50K.
·Common practice is to convert this to binary(1 for>50K,O for<=50K).
Feature List
·age(numeric)
·workclass (categorical:Private,Self-emp,Government,etc.)
·fnlwgt(numeric:“final weight,”representing how many people in the US population each record represents)
·education (categorical:Bachelors,HS-grad,etc.)
·education_num (numeric:1-16,encoded years of education)
·marital_status(categorical)
·occupation (categorical)
·relationship(categorical:Husband,Wife,Not-in-family,etc.)
·race (categorical)
·sex(categorical:Male/Female)
·capital_gain(numeric)
·capital_loss(numeric)
·hours_per_week (numeric)
·native_country(categorical)
·income (target:>50K/<=50K)
Task Overview
Data Acquisition &Understanding(Code provided)
·Download the dataset (e.g.,adult.data from the UCI Repository or Kaggle).
·Familiarize yourself with the 14 features and the target column (>50K/<=50K).
Data Cleaning
·Import the dataset into a DataFrame (Code provided)
·Identify and handle missing values (often represented by"?").Decide whether to drop or impute those rows( 0.25 points). Feature Engineering &Encoding
·Convert the target (income)to a binary numeric:1 if>50K,0 if<=50K(0.25 points).
·Encode categorical columns appropriately(e.g.,workclass,education,marital_status):(0.5 points)
■ One-hot encoding(dummy variables)or label encoding.
·Consider dropping high-cardinality or rarely occurring categories,or grouping them.
Data Splitting:Split into train and test sets(0.5 points)
Model Training:Select a suitable model and appropriate columns to train the model.(0.5 points)
Evaluation:( 0.5 points)
·Generate predictions on the test set and compute classification metrics:
■ Accuracy
■ Precision,Recall,F1-score
■ Confusion matrix Prediction:Make an imaginary person,use the model to predict whether the person's income will be above 50K(0.5 points).
|
#If you have not installed the UCI Machine Learning Repo module,un-comment the next line and install it.
#!pip install ucimlrepo
|
|
#This is the part you download the dataset and convert it to a pandas data frame.
from ucimlrepo import fetch_ucirepo
import pandas as pd
import numpy as np
adult =fetch ucirepo(id=2) A=adult.data.features
B=adult.data.targets
df=pd.concat([A,B],axis=1) df
|