This is a method of identifying tracks that are "spiking" on a given label's catalog on a given day.

How we identify a “spike”:

On a given day (most recent day of available Spotify data):
1. Look at a label’s top 10% of tracks in terms of Spotify streaming
2. From that group, drop any tracks with <200 Spotify streams
3. For the remaining tracks, get the:
    - Average daily Spotify streams over the past 90 days
    - Standard deviation of Spotify streams over the past 90 days
    - Spotify streams on most recent day
4. Use the above data to calculate a Z-score:
Z = (streams_today - average_streams) / (standard_deviation_streams)
5. Any track with Z > 4.5 is “spiking”

historical.sql gets:
- TRACKID
- TRACKNAME
- ARTISTNAME
- PERCENT_RANK
- AVG
- STD

lw.sql is used to generate trend lines for the past week for spiking tracks (to visualize the spike)
- TRACKNAME
- ARTISTNAME
- DAY (gets last 7 days)
- TRACKID
- PERCENT_RANK
- SUMS (Spotify streams)

placements.sql is used to get any recent playlist placements for spiking tracks
- TRACK
- ARTIST
- pl.NAME
- LINK
- DATE_ADDED
- STREAMS (from playlist)

today.sql gets the most recent day's data:
- TRACKNAME
- TRACKID
- ARTISTNAME
- PERCNT_RANK
- ISRCID
- TODAY (streams on most recent day)

Finally, spike_detector_script.py takes all the data gathered from the above queries, puts it into Pandas dataframes, and does the necessary calculations to get Z-score, print spiking tracks, print placements, and print trend graphs.