# How to replace pandas-td

Pandas TD is no longer maintained, we recommend you to switch to **pytd** by following the porting guide below.

**pytd** offers [pandas-td](https://github.com/treasure-data/pandas-td) compatible functions that provide the same functionalities more efficiently.

## Installing

If you have not already installed pytd then do so.


```shell
pip install pytd
```

## Configuration

Change from `pandas_td` to `pytd.pandas_td` in your import statements and when you load the ipython extensions.

**Before**


```python
import pandas_td as td
```


```shell
In [1]: %%load_ext pandas_td.ipython
```

**After**


```python
import pytd.pandas_td as td
```


```ipython
In [1]: %%load_ext pytd.pandas_td.ipython
```

After these configuration changes, all `pandas_td` code should keep running correctly with `pytd`.

Info
There is a known difference with `pandas_td.to_td` function for type conversion. Since `pytd.writer.BulkImportWriter`, default writer in pytd, uses CSV as an intermediate file before uploading a table, column type may change via `pandas.read_csv`. To respect column type as much as possible, you need to pass `fmt="msgpack"` argument to `to_td` function.

For more detail, see `fmt` option of `pytd.pandas_td.to_td()`.