Double median filter and weighted clustering

Two things to try:

  • First filter out forecasts which are 2SD away from the median
  • Then filter out forecasts which are 1 SD away from the median

and:

  • Plot forecasts on Y axis being accuracy and X axis being level
  • Use k-means clustering to find 2 clusters
  • Pick the cluster whose sum of weights is greatest
  • Then single or double-filter that cluster as above
Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s