from app import *
from mlxtend.regressor import LinearRegression

def get_predict(df):
    feature_names=df.columns.to_list()[1:]
    target_name=df.columns.to_list()[1]
    # use 60 days to predict
    window_size=60
    x=[]
    y=[]
    for index in range(len(df)):
        if index >= window_size:
            features=np.array([])
            for feature_name in feature_names:
                feature=df.loc[index-window_size:index-1, feature_name].to_numpy()
                features=np.append(features,feature)
            x.append(features)
            target=df.loc[index, target_name]
            y.append(target)
    x=np.array(x)
    y=np.array(y)

    # train
    model = LinearRegression()
    model.fit(x, y)

    # current
    days=60
    plot_df=df.copy()
    plot_df["Type"] = 'Current'
    # predict
    for i in range(days):
        index=len(plot_df)-window_size
        predict_x=np.array([])
        for feature_name in feature_names:
            feature=plot_df[feature_name].iloc[index:].to_numpy()
            predict_x=np.append(predict_x,feature)
        # transform
        predict_x=np.array([predict_x])
        
        # predict
        predict_y = model.predict(predict_x)

        predict_date=plot_df['Date'].iloc[-1]+pd.Timedelta(1, unit='D')
        predict = pd.DataFrame({'Date':predict_date, feature_names[1]:plot_df[feature_names[1]].mean(), feature_names[0]:predict_y, 'Type':'Predict'})
        plot_df = pd.concat([plot_df,predict])
        plot_df.reset_index(drop=True, inplace=True) 
        plot_df.sort_values(['Date'],inplace=True)
    return(plot_df)

# layout
ml_list=['Predict Covid-19 Fatal Rate', 'Predict CBM']
layout = html.Div([
    dcc.Dropdown(
        id='input-ml',
        options=[{'label':a, 'value':i} for i, a in enumerate(ml_list)]),
    dcc.Graph(id='output-predict_fatal_rate')
    ])

# use default option
@app.callback(Output('input-ml', 'value'),
        Input('data-machine_learning', 'data'))
def get_ml(data):
    return(0)

@app.callback(Output('output-predict_fatal_rate', 'figure'),
        Input('input-ml', 'value'),
        State('data-machine_learning', 'data'))
def update_output(selected_ml, data):
    to_df(data)
    if selected_ml==0:
        covid_world_wide_fatal_rate=data['covid_world_wide_fatal_rate']
        predict_fatal_rate=get_predict(covid_world_wide_fatal_rate)
        fig_predict_fatal_rate = px.line(predict_fatal_rate, x="Date", y='Death percentage', color='Type', title='Fatal Rate Prediction')
        fig_predict_fatal_rate.update_yaxes(tickformat=".2%")
        update_theme(fig_predict_fatal_rate)
        return(fig_predict_fatal_rate)
    elif selected_ml==1:
        cbm_ctn=data['cbm_ctn']
        predict_cbm=get_predict(cbm_ctn)
        fig_predict_cbm = px.line(predict_cbm, x="Date", y='Daily CBM', color='Type', title='CBM Prediction')
        update_theme(fig_predict_cbm)
        return(fig_predict_cbm)
    else:
        raise(PreventUpdate)