STAT5243-Project-2 -- Web App Development

This project is an interactive Shiny web application for dataset upload, preprocessing, feature engineering, and exploratory data analysis.

App Workflow

The application follows a 4-step data analysis pipeline:

• Dataset Upload & Preview
• Data Cleaning
• Feature Engineering
• Exploratory Data Analysis (EDA)

Step 1: Dataset Upload & Preview

What to Do

• Choose:
• Upload file, or
• Use sample dataset (iris / mtcars)
• If uploading:
• Click Upload a dataset
• Supported formats: .csv .xlsx / .xls .rds .json

What You Get

• Dataset structure:
• Number of rows and columns
• Column names and types
• Missing values per column
• Data preview:
• First 10 rows of the dataset

Step 2: Data Cleaning

Available Operations (can be selected from the side bar)

• Remove Duplicate Rows

  • 1. Removes exact duplicate observations

• Handle Missing Values

  • 1. Choose one method from the drop down menu Missing value method:
    • Remove rows with missing values
    • Mean imputation (numeric only)

• Handle outliers

  • 1. Choose one method from the drop down menu Outlier method:
    • Cap outliers using IQR
    • Remove rows with outliers

• Scale Numeric Variables (Optional)

  • 1. Choose one method from the drop down menu Scaling method:
    • Standardization (z-score)
    • Min-Max scaling

How to Use

• Select cleaning options in the sidebar
• Click Apply Data Cleaning
• Go to Data Cleaning tab panel

Outputs

• Cleaned dataset preview
• Cleaning log (step-by-step actions)

Download

user can choose to download cleaned dataset and cleaning log as csv files by clicking Download Cleaned Dataset and Download Cleaning Log at the side bar

• cleaned_dataset.csv
• cleaning_log.csv

Step 3: Feature Engineering

Supported Methods

user can choose which feature engineering type to use

• Single-variable Transformations
  • 1. Choose transformation type through drop down menu Choose transformation:
    • Log: log(x + 1)
    • Square: x²
    • Square root: √x
• Two-variable Interaction Features
  • 1. Choose operation type through drop down menu Choose operation:
    • Multiply: x * y
    • Divide: x / y (safe division)
    • Add: x + y
    • Subtract: x - y
• Binning
  • 1. Select the number of bins in Number of bins:

How to Use

• Select feature engineering type
• Choose variables
• Click Apply Feature Engineering
• Check results in Feature Engineering tab panel

Key Features

• Only numeric variables are selectable
• Prevents invalid operations (e.g., division by zero)
• Automatically generates unique feature names
• Updates dataset instantly

Outputs

• Updated dataset with new features
• Feature engineering log

Download

user can choose to download engineered dataset and feature log as csv files by clicking Download Engineered Dataset and Download Feature Log at the side bar

• engineered_dataset.csv
• feature_log.csv

Step 4: Exploratory Data Analysis (EDA)

Choosing Dataset

At the Exploratory Data Analysis tab panel, under EDA Controls, user can choose dataset by selecting either:

• Current engineered dataset
• Cleaned dataset
• Original raw dataset

Choosing Plot Type (plot will be shown under 'EDA Plot'.)

Under EDA Controls, user can choose the plot type they want to generate throught the drop down menu Choose plot type:

• Histogram
  • 1. Select the numeric variable through drop down menu Select numeric variable:
• Boxplot
  • 1. Select the numeric variable through drop down menu Select numeric variable:
  • 2. Select the categorical variable through drop down menu Select categorical variable:
• Bar Chart
  • 1. Select the categorical variable through drop down menu Select categorical variable:
• Scatter Plot
  • 1. Select the numeric variable through drop down menu Select numeric variable:
  • 2. Select the X variable through drop down menu Select X variable:
  • 3. Select the Y variable through drop down menu Select Y variable:

How to Use

• Select dataset
• Choose plot type
• Choose primary variable
• (Optional) Choose secondary variable
• View plot and summary on the right side of the tab panel

Outputs

• Dataset Summary
• EDA Plot
• Missing Values by Column
• Numeric Summary
• Correlation Analysis
• Correlation Matrix
• Categorical Summary

Team Members

• Carrie Yan Yin Feng – Data upload and preview
• Shuzhi Yang – Data cleaning
• Haoyun Tong – Feature engineering
• Yolanda He – EDA