diff --git a/speeds2.py b/speeds2.py
new file mode 100644
index 0000000..10d156a
--- /dev/null
+++ b/speeds2.py
@@ -0,0 +1,40 @@
+import gpxpy
+from os.path import expanduser
+from math import sqrt
+from shapely.geometry import Point
+import geopandas
+import numpy
+from matplotlib import pyplot
+
+with open(expanduser('~/Desktop/2016-06-14 19_13_54.gpx'), 'r') as gpx_file:
+  gpx = gpxpy.parse(gpx_file)
+
+gpx.tracks[0].segments[0].points[0].time
+gpx.tracks[0].segments[0].points[0].latitude
+gpx.tracks[0].segments[0].points[0].longitude
+
+speeds = []
+for i in range(0, len(gpx.tracks[0].segments[0].points) - 1):
+  deltaLatitude = gpx.tracks[0].segments[0].points[i].latitude - gpx.tracks[0].segments[0].points[i - 1].latitude
+  deltaLongitude = gpx.tracks[0].segments[0].points[i].longitude - gpx.tracks[0].segments[0].points[i - 1].longitude
+  speed = sqrt(deltaLatitude**2 + deltaLongitude**2) * 69.04799998422561 * 3600 # not sure if this is the right conversion
+  midLatitude = (gpx.tracks[0].segments[0].points[i].latitude + gpx.tracks[0].segments[0].points[i - 1].latitude) / 2 
+  midLongitude = (gpx.tracks[0].segments[0].points[i].longitude + gpx.tracks[0].segments[0].points[i - 1].longitude) /2
+  speeds.append((Point(midLongitude, midLatitude), midLatitude, midLongitude, speed))
+
+speedsDF = geopandas.GeoDataFrame(speeds, columns=('geometry', 'latitude', 'longitude', 'speed'))
+
+# from http://stackoverflow.com/questions/25526682/functions-to-smooth-a-time-series-with-known-dips
+filtered = speedsDF.speed.copy()
+dm = speedsDF.speed.rolling(window=20,center=True).median()
+df = sorted(numpy.abs(speedsDF.speed - dm).dropna(), reverse=True)
+cutoff = df[len(df) // 20] 
+filtered[numpy.abs(speedsDF.speed - dm) > cutoff] = numpy.nan
+filtered[0] = numpy.nan
+speedsDF['filtered'] = filtered
+speedsDF.dropna(subset=['filtered'], inplace=True)
+
+f, ax = pyplot.subplots(2, sharex=True)
+speedsDF.plot(ax=ax[0], column='filtered', cmap='OrRd', markersize=10, marker='.')
+ax[1].plot(speedsDF.longitude, speedsDF.filtered)
+pyplot.show()