Pattern Discovery in Time Series Data Using Python Script and MS Excel

Pattern discovery is a novel approach to time series data. Even though many attempts are made, some innovative ideas for pattern discovery may be proposed. The purpose of this paper is to present an approach for extracting Yahoo finance data, converting it into MS Excel data, and discovering patterns in MS Excel. The main approach contains two steps: first, using a Python script for Yahoo finance data extraction and saving it as an MS Excel spreadsheet (time series data for a specific sector for a certain period), and second, using simple Excel formulas for pattern discovery. A standard count of pattern occurrences is calculated according to the length of the pattern. Patterns with different sizes are checked. Each pattern is checked – its appearances count in the time series compared to the standard count of occurrences. If the pattern occurrences count is greater than the standard occurrences count, it may be concluded that the pattern discovery method is successful. The paper presents an original (novel) approach to pattern discovery. The approach is limited to a single sector, but it can be applied to pattern discovery in any business sector. The paper is useful for practitioners who seek advanced forecasting techniques based on pattern discovery. The performance testing of different patterns is done by visualizing their occurrence count in the time series.