-
plot_all_decel.py
: plots all calibrated accelerations found in a folder -
plot_trips_selection.py
allows to compare a selection of trips, both for calibrated/uncalibrated and raw/filtered data -
plot_FFT_single_trip.py
: plots the FFT of calibrated accel of a single trip
-
find_constant_accel.py
was the first attempt to identify time windows where accel is constant. Leads to false positives (see ep.2 : https://www.linkedin.com/pulse/data-analysis-paris-metro-ep2-eric-cabrol-tq7ye/) -
find_jolts.py
: second attempt, using "jolt-back" peaks => seems better (see ep.3 : https://www.linkedin.com/pulse/data-analysis-paris-metro-ep3-eric-cabrol-lynie) A filetimestamps.txt
is generated in each folder. It still requires a manual validation, then the file must be saved astimestamps_valid.txt
. -
find_jolts_all.py
: detect jolts in all recordings of a given directory => in fact there are as many difficulties to identify the stops, so I switched back to the constant accel solution -
find_constant_accel_all.py
: asks for confirmation after each trip, which allows to modify the timestamps file before validating
- previously done in
find_stops_from_trip_name.py
, now inmetro.py
module
cut_trips.py
: could be merged withcheck_timestamps.py
metro.py
: contains a class named Trip to retrieve all useful information from the trip name. Shall I also manage quantitative data with this class (eg sampling frequency) ?
NB : zip files exported from SensorLogger must be manually downloaded from Google Drive into zip
folder
unzip_recordings.py
allows to unzip recordings indata
remove_total_accel.py
removesTotalAcceleration.csv
files, because they contain the same data asAccelerometerUncalibrated.csv