Corona-Virus-Prediction

Time series analysis with Corona Virus Daily Data and ARIMA models

ARIMA(Autoregressive integrated moving average) 란?

In statistics and econometrics, and in particular in time series analysis, an autoregressive integrated moving average (ARIMA) model is a generalization of an autoregressive moving average (ARMA) model. Both of these models are fitted to time series data either to better understand the data or to predict future points in the series (forecasting). WIKIPEDIA

Example

1. Use getData.py to bring ./data/data.csv Corona virus daily status Dataset

2. Show a Data Graph

data = getData.read_csv('list')

def ShowGraph(data):
  df = pd.DataFrame(data, columns = ['date' , 'value'])
  df.date = pd.to_datetime(df.date)
  df.value = pd.to_numeric(df['value'])
  df = df.set_index('date')
  g = df['value'].plot(title="Corona-Virus Daily data")
  plt = g.get_figure()
  ...

3. ACF & PAF Calculation and Show a graphs

def Calc_ACF_PAF():
  ...
  plot_acf(data)
  plot_pacf(data)
  ...

Left : ACF / Right : PACF

4. Forecast using ARIMA models

def ARIMA():
  ...
  order = (0,1,1)
  model = statsmodels.tsa.arima_model.ARIMA(series, order, freq='D')
  model_fit = model.fit(trend='c',full_output=True, disp=10)
  ...

5. Show a graph

def ARIMA():
  ...
  plt = model_fit.plot_predict()
  Number_to_predict = 3
  fore = model_fit.forecast(steps=Number_to_predict)
  ...

Orange Line : Origin / Blue Line : Forecast

6. Result

for x in fore[0]:
        p_time = time + timedelta(days=Number_to_predict)
        p_time = p_time.strftime('%Y-%m-%d')
        print("{0} : {1}".format(p_time, x))
        Number_to_predict += 1

# Output
2020-03-09 : 6057.29830482379
2020-03-10 : 6196.215619909736
2020-03-11 : 6335.132934995682

Differences between actual data and forecast results

날짜	Predict	Actuality	Difference
2020-03-09	6057	7382	1325
2020-03-10	6196	7513	1317
2020-03-11	6335	7755	1420

Execution / Test Environment

• Window 10
• Python 3.6