Forest Cover Type

Project is created by Adziareika Aliaksandr

LinkedIn
GitHub project

Description

There are two main endpoints to get the data:

• Predict
• Get the example input

All the requests are provided with postman as well.
All the models are saved in the models directory.

Tasks

Task 2

I decided to base Heuristics on Elevation. The range of values is divided into 7 about equal parts and depending on it the class is taken.

Task 3

Machine Learning Models. I decided to use Logistic Regression and SGD Classifier. The first is pretty simple and basic, which became a reason. And SGD is chosen because the data set is quite huge and we solving the issues of classification.

Task 4

Neural Networks. RandomSearch was used to adjust hyperparameters. And a few of Dense, Dropout layers have been added.

Task 5

Evaluation. Some Confusion matricies were logged out, accuracy metric. History charts for NN are also shown.

Task 6

REST API it touchable and playable :)

Endpoints

Predict

POST
/predict

Body:

{
    "model_name": "heuristics | nn | logistic | sgd",
    "input": [[...]]
}

Get the example input

GET
/input/example

Response:

{
    "input": [[2596, 51, 3, 258, 0, 510, 221, 232, 148, 6279, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0,
    0, 0, 0, 0, 0]]
}