Daily Steps Time Series

My Fitbit finally gave up after almost 5 years. After reading the series Analyzing Time Series Data, I really want to apply some of these techniques to my daily steps data.

Filtering and Smoothing

With 1761 data points in my daily steps data, it can be difficult to see the trend. To reduce the complexity, we can utilize a brush filter to zoom into a date interval of interest. To reduce the noise, we can use the moving average method to smooth the data.

Brush to filter the dates
Daily (no smoothing)
Weekly moving average
28-day period moving average

Seasonal Patterns

By creating additional date features (day of the week, weekend vs weekday, month of the year, quarter of the year), we can group the data points by these date features and look for any seasonal patterns that's not easy to spot.

Each barcode chart inside the grid consists of a rectangle background colored by a chosen statistic feature and lines representing one day's step count.

Horizontally grouped by
Vertically grouped by
Colored by