df_full['increase'] =(df_full['2018-04'] - df_full['1996-04'])  #calculate price change and the percentage of the increase
df_full['increase_rate'] = df_full['increase']/df_full['1996-04'] 
df_full['criteria']= df_full['increase']*df_full['increase_rate']

def drop_cols(df):
    ts=df
    for col in ts.columns:
        if col in ('RegionID', 'CountyName','Metro','SizeRank','State','increase','increase_rate','criteria'):
            ts=ts.drop(col, axis=1)
    return ts
        

def convert_to_ts (df):   
    ts=drop_cols(df)  #drop columns other than price data
    ts=ts.T     # Transpose the dataframe
    ts.index=pd.to_datetime(ts.index)  #change index type to timestamp
    ts=ts.astype(int)  
    return ts

from statsmodels.graphics.tsaplots import plot_acf, plot_pacf
def rolmean_rolstd_plot(df, zipcode):
    # check rolmean , rolstd plots for trend
    rolmean = df.rolling(window = 4).mean()
    rolstd = df.rolling(window = 4).std()
    fig = plt.figure(figsize=(12,4))
    orig = plt.plot(df, color='blue',label='Original')
    mean = plt.plot(rolmean, color='red', label='Rolling Mean')
    std = plt.plot(rolstd, color='green', label='Rolling Std')
    plt.legend(loc='best')
    plt.title(zipcode+' Parameters: Rolling Mean & Standard Deviation')
    plt.show(block=False)
    
    # check first degree diff() plot for detrending
    ts_lag_1 = df.diff(1).plot(legend=False, label='1 lag difference',linewidth=1, c='r')
    plt.xlabel('1 lag difference')
    rcParams['figure.figsize'] = 10,1
    
    #ckeck Autocorrelation , Partial Autocorrelation plots
    ts_acf_plot = plot_acf(df, title='ACF')
    ts_pacf_plot = plot_pacf(df, title='PACF')

for col in ts_top10_zip.columns:
     rolmean_rolstd_plot(ts[col],str(col))

    for i in ts.columns:
    X = ts[i]
    result = adfuller(X)
    print(i)
    print('ADF Statistic: %f' % result[0])
    print('p-value: %f' % result[1])
    print('Critical Values:')
    for key, value in result[4].items():
        print('\t%s: %.3f' % (key, value))

from sklearn.metrics import mean_squared_error
pdq_new=(3,1,1)
pdqs_new=(3,1,1,12)

for i in zipcodes:
    X = ts[i]
    size = int(len(X) * 0.66)
    train, test = X[0:size], X[size:len(X)]
    history = [x for x in train]
    predictions = list()

    for t in range(len(test)):
        model = sm.tsa.statespace.SARIMAX(history,
                                        order=pdq_new,
                                        seasonal_order=pdqs_new,
                                        enforce_stationarity=False,
                                        enforce_invertibility=False)
        model_fit = model.fit()
        output = model_fit.forecast()
        yhat = output[0]
        predictions.append(yhat)
        obs = test[t]
        history.append(obs)
        #print('predicted=%f, expected=%f' % (yhat, obs))

    error = np.sqrt(mean_squared_error(test, predictions))
    print("Zipcode :", i )
    print('Test RMSE: %.3f' % error)
    print('Standart Deviation:' , np.std(test))
    print(model_fit.summary())
    
    plt.figure(figsize=(14,5))
    #plt.plot(train, label='Train')
    plt.plot(test, color='red', label="Test")
    plt.plot(test.index, predictions, color='blue', label="Predicted", alpha=0.5)
    plt.legend()
    plt.show()  

from fbprophet import Prophet as proph
def prophet_prediction(df, topzips):
    dscol = [x for x in df.ds]
    
    values = []
    
    for i in topzips:
        ycol = [x for x in df[i]]
        
        ts = pd.DataFrame()
        ts['ds'] = dscol
        ts['y'] = ycol
        
        Model = proph(interval_width=0.95)
        Model.fit(ts)
        
        future_dates = Model.make_future_dataframe(periods=36, freq='MS')
        forecast = Model.predict(future_dates)
        
        forecasted_data=forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']]
        
        increase_3_yrs = round(forecast['yhat'][300] - forecast['yhat'][264], 2)
        increase_rate_3 = round(increase_3_yrs / forecast['yhat'][264], 2)
                                                        
        values.append((i, df[i][-1], increase_3_yrs, increase_rate_3))
        
        
        Model.plot(forecast, uncertainty=True)
        plt.title(i)
        plt.ylabel('Real Estate Value')
        plt.xlabel('Year')
        Model.plot_components(forecast)
        plt.title(i)
        plt.ylabel('Real Estate Value')
        plt.xlabel('Yearly Trend');
        
        
       
        print("Zipcode :" , i)
        print("Investment($) :", df[i][-1] )
        print("********")
        print("Predicted price in 3 years :",round(forecast['yhat'][300] ))
        print('Increase rate in 3 years (%) :',  increase_rate_3 )
        print("-----------------------------------------------------------------------------")    
        
    return values , forecasted_data