Audience Segmentation

About:

A basic replica of youtube algorithm! Classifies the viewers into several groups based on their interests. Implemented with TensorFlow.

The code collects the amount of videos watched in particular category by any person from the dataset and gets the corresponding labels and classifies the type of audience for user defined person in train.py. Useful for classifying audience based on their interests and personalizing ads for each type of audience type.

Model accuracy of 95-100% is achieved while testing.

Tested with:

Operating System:

• Pop OS 20.10

Python version:

• Python 3.8.6 64-bit

Packages:

• tensorflow 2.4.1
• numpy 1.19.5
• pandas 1.2.4
• matplotlib 3.4.2
• scipy 1.6.3
• sklearn 0.24.2

Explanation:

Classification is the concept used here. Classification refers to a predictive modeling problem where a class label is predicted for a given example of input data. Here the model is trained with data of Audience where the model understands the relation between the known features (X) and their respective classes (Y). The trained model classifies the audience into various classes (Y) based on their features (X)

About the Neural Network:

A typical form of Artificial Neural Network (ANN) is used here. The layers of the Neural Network Architecture is as follows...

Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense (Dense)                (None, 14)                112       
_________________________________________________________________
dense_1 (Dense)              (None, 16)                240       
_________________________________________________________________
dense_2 (Dense)              (None, 16)                272       
_________________________________________________________________
dense_3 (Dense)              (None, 16)                272       
_________________________________________________________________
dense_4 (Dense)              (None, 7)                 119       
=================================================================
Total params: 1,015
Trainable params: 1,015
Non-trainable params: 0

Developed by:

Vigneshwar Ravichandar
Moulishankar M R

List of Classes:

1. Young adult ( if the value of 'res' in test.py is '[0]' )
2. Engineering Student ( if the value of 'res' in test.py is '[1]' )
3. Medical Student ( if the value of 'res' in test.py is '[2]' )
4. Teachers ( if the value of 'res' in test.py is '[3]' )
5. Adults ( if the value of 'res' in test.py is '[4]' )
6. Travellophilic ( if the value of 'res' in test.py is '[5]' )
7. Media Freak ( if the value of 'res' in test.py is '[6]' )

List of Features:

1. Technology
2. Politics
3. Food
4. Education
5. Media
6. Travel
7. Medicine

Execution Instructions:

Execute the following command in the terminal to run with default procedure.

python3 main.py --test=True

Command Line Arguments:

• -tr (or) --train - Used to train the Neural Network.

  • Argument type: bool
  • Parameter type: Optional
  • Default value: False

• -t (or) --test - Used to test the Neural Network with custom inputs.

  • Argument type: bool
  • Parameter type: Mandatory

• -v (or) --visualize - Used to vizualize the metrics.

  • Argument type: bool
  • Parameter type: Optional
  • Default value: False

• -req (or) --install_requirements - Used to install the required dependancies.

  • Argument type: bool
  • Parameter type: Optional
  • Default value: False

• -e (or) --epochs - Used for mentioning the number of epochs for the model.

  • Argument type: int
  • Parameter type: Optional
  • Default value: 50

• -bs (or) --batch_size - Used for mentioning the batch size for the model.

  • Argument type: int
  • Parameter type: Optional
  • Default value: 5

• -l (or) --loss - Used for mentioning the loss function for the model.

  • Argument type: str
  • Parameter type: Optional
  • Default value: "sparse_categorical_crossentropy"

• -op (or) --optimizer - Used for mentioning the optimizer for the model.

  • Argument type: str
  • Parameter type: Optional
  • Default value: "adam"

Images:

                          *Screenshot mentioning the training command*

                          *Screenshot mentioning the testing command*