Skip to main content

I/O

Shapelets supports a variety of data sources to create Dataframe-like structures to process data.

To load or save data, you need to create a sandbox on Shapelets first, you can do like:

import shapelets as sh

session = sh.sandbox()

Read Data​

1) Read from CSV files​

You can read csv files using the from_csv() function. A basic example is:

df = session.from_csv(path)

This reader function has automatic delimiter/separator detection, so you do not need to specify the delimiter!

This function has more params, like date_forma to specify the date format on date columns, and even more. You can check it out on the API Reference in from_csv()

2) Read from parquet files​

You can use from_parquet() to load parquet files. A basic example here:

df = session.from_parquet(path)

This function has more parameters, which you can find in the API Reference: from_parquet()

3) Read from Pandas DataFrames​

In the case that you have a Pandas DataFrame which you woud like to quickly load in Shapelets, you can do it so using from_pandas() function:

pandas_df = pd.DataFrame(data)
df = session.from_pandas(pandas_df)

Export/Save Data​

1) Save to a CSV file​

To save the content of a Shapelets Dataframe into a csv file, you can use the to_csv() function.

df.to_csv(filename, delimiter=",")

This function has more params, which you can find in the API Reference in to_csv()

2) Save to a parquet file​

If you want save the contents of a Shapelets Dataframe into a parquet file, you can use the to_parquet() function.

df.to_parquet(filename, delimiter=",")

This function has more params, which you can find in the API Reference in to_parquet()

3) Convert into a Pandas DataFrame​

You can load data into Shapelets, process it and convert the results directly to a Pandas Dataframe object to use it with Pandas functions or any other Python library. Here is an example:

df.to_pandas()