added the test_set analysis for checking the trends for pv_ids (#69)

* added the analysis for each pv_id trends

* added documentation for test_analysis

Author: roshnaeem

images/test_analysis_output.png

@@ -7,4 +7,18 @@ By analysing the metadata, available at [Hugging Face](https://huggingface.co/da
 1. Most of the data in the test set has a tilt angle of 30-34 degrees
 2. The maximum kwp is 4.0 & the minmum kwp is 2.25 in the test set.
 
-A detailed anaysis of the test set can be found at quartz_solar_forecast/dataset/dataset_analysis/test_set_analysis.ipynb
+A detailed anaysis of the test set can be found at quartz_solar_forecast/dataset/dataset_analysis/test_set_analysis.ipynb
+
+### `test_set_analysis_pv_id_vs_month.ipynb`
+This file uses `testset.csv`, which consists of data from 50 photovoltaic systems represented by unique `pv_id`. Each `pv_id` has 50 data points collected at times represented by `timestamp`. The dataset was analyzed to observe the distribution trends of data points during different months of the year for each PV ID.
+The following scatter plot shows the distribution of data points for each PV ID across the months of the year.
+
+![PV ID vs. Month Distribution](https://github.com/openclimatefix/Open-Source-Quartz-Solar-Forecast/blob/main/images/test_analysis_output.png?raw=true)
+
+The following observations were made from the plot:
+
+- **Distribution of Data Points**: The plot displays data points for all months across multiple PV IDs. Each dot signifies an instance of electricity generation data recorded from a PV system.
+
+- **Frequency of Data Points**: The color intensity on the scatter plot corresponds to the frequency of data points for each PV ID and month. Lighter shades represent a lower number of data points, whereas darker shades signify a higher frequency. Notably, the months of May, June, July, August, and September are marked by darker shades, indicating a higher frequency of data points compared to the rest of the year.
+
+- **Uniformity Across Months**: Data points are distributed fairly evenly across the months for each PV ID, which implies that data collection is consistent throughout the year without significant lapses.