- Python 3.5 or higher
- Numpy
- Pandas
- Matplotlib
- Statsmodels
In this project, the Dataset used is the famous Kaggle Dataset.
First we use pandas to read the CSV file. After that we perform Data Cleaning such as removing unwanted columns, identifying and filling the missing values.
This is achieved using Augmented Dickey-Fuller test which assists finding out the stationary properties in the Time series data.
The stationarity check is carried out using the plot and p-value is tested.
type of df_test: <class 'tuple'>
df_test: (-4.271198430066125, 0.00049808780753740904, 16, 508, {'1%': -3.4432882895877501, '10%': -2.5698092313534628, '5%': -2.8672462791357867}, 2060.5638228165117)
df_output:
Test Statistic -4.271198
#Lags Used 16.000000
Number of Observations Used 508.000000
dtype: float64
Test Statistic -4.271198
p-value 0.000498
#Lags Used 16.000000
Number of Observations Used 508.000000
Critical Value (1%) -3.443288
Critical Value (10%) -2.569809
Critical Value (5%) -2.867246
dtype: float64\
As the p-value is below the threshold value, thus we can reject null-hypothesis which states that the data is not stationary.
I used ARIMA model which stands for "Auto-Regressive Integrated Moving Averages".
In this case as the data is stationary therefore only AR and MA have been taken into account.
Then AIC is calculated and then I found values of p and q having lowest AIC where the p represents the number of Auto-Regressive (AR) terms and q represents the number of Moving Averages (MA) terms.
After that the results are predicted. Here is the example of results.
| Year | Temperature Forecasting (Deg C) |
|---|---|
| 2023-01-01 | -0.244643 |
| 2023-02-01 | 0.817309 |
| 2023-03-01 | 4.039086 |
| 2023-04-01 | 8.557097 |
| 2023-05-01 | 13.160331 |
| 2023-06-01 | 16.614962 |
| 2023-07-01 | 17.995051 |
| 2023-08-01 | 16.930731 |
| 2023-09-01 | 13.707325 |
| 2023-10-01 | 9.188862 |
| 2023-11-01 | 4.586472 |
| 2023-12-01 | 1.133756 |