################
###  AutoML  ###
################

AutoML is a simple, rather naive automated machine learning platform. It enable data scientists to
examine their hypothesis on a given use case.

The platform features data import, data understanding, exploratory data analysis, data pre-processing,
modelling, and deployment. Each feature is explain below in more details.

--------------
1. Data Import
--------------
Objective:
 Import data into the platform


Features:
 For this release, AutoML only accept CSV files. The data can be imported with its unique features.

Required action(s):
 - selected preferred dataset from local machine

--------------
2. Data Understanding
--------------
Objective:
 This stage helps data analysts/scientists to study their data more closely by giving the mentioned information.

Features:
 Once the data is uploaded, AutoML shifts the user to the data understanding page where it display the following:
    - data sample
    - data shape
    - feature type and sum of null values
    - data description
    - feature box plot for numeric columns, and histogram for categorical columns

Required action(s):
 - Unique columns are removed at this stage
 - selection of target feature
 - option to keep/remove outliers
 - option to keep/deal-with null value

--------------
3. EDA
--------------
Objective:
  This stage provides analysis of the data offering a more concise view on the data to data analysts/scientists

Features:
 One the user has selected the target feature and their preferences, AutoML shifts the user to the exploratory data analysis page where it display the following:
    - converting categorical data into numerical form
    - dealing with outliers
    - dealing with null values
    - data distribution per feature
    - Feature importance
    - Feature correlation

Required action(s):
- selection of modelling features

--------------
4. Data pre-processing
--------------
Objective:
 In this stage, the selected features are preprocessed for modelling

Features:
 Once the features are selected, they will be preprocessed. Preprocessing includes:
    - defining independent variables
    - defining dependant variable
    - apply one hot encoding on the independent variables
    - splitting the data into training and testing sets
    - scaling independent variables

Required action(s):
- N/A

--------------
5. Data Processing/Modelling
--------------
Objective:
 This stage models the data and provide information for each model.

Features:
 Data modelling includes:
    - grid search for the optimal hyper-parameter
    - modelling the data
    - providing the baseline
    - providing a classification report
    - recommendation
    - save model

Required action(s):
- option to select model to be deployed.

--------------
6. Deployment
--------------
Objective:
 Deploy model to be used for prediction - inference

Features:
 - load model
 - prediction

Required action(s):
- Post request to model API