What’s the best pandas to dict method to achieve the following output format:
{'p': [1, 3, 2], 'q': [4, 3, 2], 'r': [4, 0, 9]}
Over time, I’ve found a pretty clean way to convert a DataFrame into a dictionary. My go-to method has always been this one-liner:
df.set_index('ID').T.to_dict('list')
It sets the first column as the index and transposes the DataFrame, so you get the remaining columns as lists. It’s been my reliable pandas-to-dict trick when reshaping data for APIs. You don’t need to manually loop through each row, and it’s super efficient.
That’s a neat approach! But in certain cases, especially when I need to tweak or clean the data before storing it, I prefer to build the dictionary manually. Something like this works great:
result = {row[0]: list(row[1:]) for row in df.values}
This way, I get full flexibility, perfect when I want to clean or transform the data mid-conversion. It gives me that extra control since it uses NumPy-like access directly to the DataFrame. I find it super helpful when the structure isn’t as straightforward.
Great ideas! I’ve used a similar technique but with a bit of added flexibility. Sometimes I need to apply custom logic to each row, so I use something like this:
df.set_index('ID').apply(lambda row: row.tolist(), axis=1).to_dict()
This approach still gives you the same dictionary output but allows you to adjust the logic easily within the lambda. It’s perfect for when you need to tweak things a bit before converting the DataFrame into a dictionary. It’s definitely readable and efficient when the data isn’t entirely clean or needs specific formatting before you convert it.