4
$\begingroup$

I have data that looks like:

 1495573445.162, 0, 0.021973, 0.012283, -0.995468, 1
 1495573445.172, 0, 0.021072, 0.013779, -0.994308, 1
 1495573445.182, 0, 0.020157, 0.015717, -0.995575, 1
 1495573445.192, 0, 0.017883, 0.012756, -0.993927, 1
 1495573445.202, 0, 0.021194, 0.012161, -0.994705, 1
 1495573445.212, 0, 0.019638, 0.013718, -0.994019, 1
 1495573445.222, 0, 0.019440, 0.010803, -0.994476, 1
 1495573445.232, 0, 0.018112, 0.010849, -0.993073, 1
 1495573445.242, 0, 0.020157, 0.011154, -0.994644, 1
 1495573445.252, 0, 0.020340, 0.010040, -0.995804, 1
 1495573445.262, 0, 0.017792, 0.009857, -0.996078, 1
 1495573445.272, 0, 0.020538, 0.010239, -0.994858, 1

This is accelerometer data where the data frame columns are labeled "Time stamp", "Time skipped", "x", "y", "z", and "label" with the index set to "Time stamp".

The sampling rate is around 100Hz. How should I create a sliding window in this case?

I came up with this:

def sliding_window(data, window_size, step_size):
    data = pd.rolling_window(data, window_size)
    data = data[step_size - 1 :: step_size]
    print data
    return data

I doubt this is the correct answer, and I don't know what to set window_size and step_size given that I have a 100Hz sampling rate.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
4
$\begingroup$

You can use the built-in Pandas functions to do it:

df["Time stamp"] = pd.to_datetime(df["Time stamp"]) # Convert column type to be datetime
indexed_df = df.set_index(["Time stamp"])           # Create a datetime index
indexed_df.rolling(100)                             # Create rolling windows
indexed_df.rolling(100).mean()                      # Then apply functions to rolling windows

This code is not 100% correct because the datetime conversion is not correct but it should help you get started.

Improve this answer
$\endgroup$
4
  • $\begingroup$ data is passed in from a function that does df = pd.read_csv(filename, header=None, names=['timestamp', 'time skipped', 'x', 'y', 'z', 'label']).set_index('timestamp') followed by df.assign(dx=df.x.diff(), dy=df.y.diff(), dz=df.z.diff()) $\endgroup$ Commented Jul 18, 2017 at 18:21
  • $\begingroup$ With that said, do I need the first two lines? $\endgroup$ Commented Jul 18, 2017 at 18:22
  • $\begingroup$ "timestamp" column needs to be cast as datetime type to then later leverage rolling method. Pandas might automagically do that for you. I would be explicit about datetime casting. It is tricky. $\endgroup$ Commented Jul 18, 2017 at 18:34
  • $\begingroup$ When I try to do df['timestamp'] = pd.to_datetime(df['timestamp']) after df = pd.read_csv(filename, header=None, names=['timestamp', 'time skipped', 'x', 'y', 'z', 'label']).set_index('timestamp') I get KeyError: 'timestamp' $\endgroup$ Commented Jul 18, 2017 at 19:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.