Time series analysis of historical sea level records for every gauge along River Thames, 1911-1995. Using machine learning, an isolation forest model was appleid to identify anomalies in the water level data. Results were plotted using Bokeh.
Methods:
- Wrote function to import and load 13 datasets, one for each tidal gauge
- Performed manipulation, cleaning, and merging to prepare datasets for analysis
- Defined interquartile range function and used it to calculate summary statistics
- Calculated ratios based on high tide and low tide statistics
- Wrote for loops to generate line plots displaying monthly average water levels (with confidence intervals) across all years for each gauge location
- Wrote for loops to generate boxplots and inspect water level distributions for each gauge location
- Fit an isolation forest model to identify anomalies in the water level data and plotted the results using Bokeh
Data sources: