import/read events from a CSV file

[Argubi]

Hi, I recently discoverd ktk and it looks really promising. I followed your instructional but could not for the life of me discover how to read events, that are already part of a .csv file that I import as a dataframa and afterwards transfer into a Timeseries Object.
I have EMgData which is exported as a CSV file and Gait events (heelstrike/toe off) are stored in one (or two) columns in the csv export as "On" and "Off" (strings). I would like to import these as events. As of now they are part of the array which contains the emg data.
Am I missing something?
All the best and thank you for your response!

[felixchenier]

Hi @Argubi
I'm sure we'll make it work. Each csv file is different, so there could not be a standard way to read it. I think you are already well started by reading the data in a DataFrame. For the events, the trick would be to parse the events using a for loop that calls TimeSeries.add_event. Maybe you could post the first few lines of your CSV file so that I can understand how it is structured, and the code you tried so far? We can then start from this.

[Argubi]

Example.csv
Hi, thank you for the super fast response. The CSV file looks like this...
Greetings

[Argubi]

My very first approach as workaround to see if I can create events according to the file (without a loop) looked like this:

import kineticstoolkit.lab as ktk
import pandas as pd

df =    pd.read_csv(("Example.csv"),index_col="Time (s):")

on_indices = df[df.iloc[:, -1] == "On"].index
off_indices = df[df.iloc[:, -1] == "Off"].index

ts = ktk.TimeSeries.from_dataframe(df)

ts = ts.add_event(on_indices[0], "On")
ts = ts.add_event(off_indices[0], "Off") 

[felixchenier]

@Argubi I think you managed very well. Your code works almost perfectly on my side, I had to make two small modifications:

1. Specify that the separator is ; in read_csv.
2. Convert the data to numeric in the DataFrame. For some reasons, some of your data are not numbers, and therefore these columns are read as strings instead of numbers. I think your process for generating the CSV has a bug, since we see dates (?) instead of data at some places in the file. This may explain the numerous erroneous data in the plot.

On my side, running this code

import kineticstoolkit.lab as ktk
import pandas as pd

df = pd.read_csv(("Example.csv"), index_col="Time (s):", sep=";")

on_indices = df[df.iloc[:, -1] == "On"].index
off_indices = df[df.iloc[:, -1] == "Off"].index

for column in df.columns:
    df[column] = pd.to_numeric(df[column], 'coerce')

ts = ktk.TimeSeries.from_dataframe(df)

ts = ts.add_event(on_indices[0], "On")
ts = ts.add_event(off_indices[0], "Off")

ts.plot()

gives this figure: