Human Activity Recognizer Using Smartphone

Exploratory Data Analysis

Loading the datasets
Cleaning the data
- Check for Duplicates
- Checking for NaN/null values
- Checking for data imbalance
- Analysis of the activity feature to understand how many types of activites have been performed
- Cleaning the feature names for an easy understanding
Saving the new dataframes with cleaned feature names in csv files¶
Exploratory Data Analysis
- Visualization of the Stationary and Moving activities
- Maximizing the Static and Dynamic activities plots
- Visualization of the magnitude of acceleration data captured by the accelerometer sensor of the smartphone
- Visualizing the position of GravityAccelerationComponants data captured by the gyroscopic sensor of the smartphone
- Visualizations for feature distribution in space.
- Visualizations with stripplot
Apply t-sne on the data

ML CLASSIFIERS approach
- Importing libraries
- Loading the cleaned datasets
- Defining a function to plot the confusion matrix
- Defining a generic function to run any model specified
- Defining a function to print the gridsearch Attributes for Hypertuning the model
- Logistic Regression with GridSearchCV
- Linear SVC with GridSearch
- Decision Trees with GridSearchCV
- Comparing all models
- Getting predictions using the Hypertuned Logistic regression model
- Getting predictions using the Hypertuned Linear SVC model
Artificial Neural Network(ANN) approach
- Loading datsets
- Pre-processing and data preparation to feed data into Artificial Neural Network.
- Feature Scaling
- Using Principle Component Analysis(PCA) approach for Dimensionality Reduction = Building the ANN Model = Saving the model

UppuluriKalyani / ML-Nexus

Human Activity Recognizer Using Smartphone #298