The head of the SAIT FIITA Department of VNTU continues to share his experience in the field of artificial intelligence and Data Science in Python. He published a new (already the 25th!) video “Exploratory Data Analysis (EDA) using the example of the Bitcoin exchange rate” on his YouTube channel “AI-ML-DS Training Course in Python”: https://www.youtube.com/watch?v=ZEk966jtmjg…
The previously developed notebook for analyzing and forecasting the Bitcoin price, which can easily be used to forecast other cryptocurrencies (and other time series), has been improved:

– Divided into two: separately - analysis and separately - forecasting, since the laptop was too difficult to perceive and overloaded with graphs. Both are already adapted to work with the current date, and not fixed only until 2021.

– The “Crypto – BTC: Advanced EDA” analysis notebook (https://www.kaggle.com/code/vbmokin/crypto-btc-advanced-eda) has been supplemented with a new section (it is described in the new video), more on that later.

– The forecasting notebook “Crypto – BTC : Analysis & Forecasting” (https://www.kaggle.com/…/crypto-btc-analysis-forecasting) has been optimized by removing the manual method of determining the parameters of the ARIMA model. Experience has shown that the automatic option works quite well too, although, as a rule, the Prophet model or one of the multivariate models is the best - ARIMA models either work worse or need improvement.

In the new data analysis notebook, various graphs are built, the time series is checked for stationarity, seasonality, the dates of anomalous values of the series are analyzed, comparisons are made with COVID-19 mortality data in the USA, etc., but the last section is new - the use of libraries for automatic data analysis over different time periods: using the Sweetviz and AutoViz libraries. The video describes the features of their use and demonstrates the result with an example.

The material will be useful both for those who are studying the automation of graph construction in Python, and for those who are currently writing their master's thesis at FIITA VNTU, where, as a rule, more analytics are required.