fix: redundant writes to dataframe #1105

[X-TRON404]

I have submitted a pull request to address issue #1105. This entails first establishing a data structure, making modifications to it, and then generating a dataframe—all in a single instance.

[X-TRON404]

I've made an update to the section of the code where we initially create the features DataFrame and then modify it by accessing its elements repeatedly. In the improved version, I'm now constructing the features using a native Python data structure first, and then transferring these features to the DataFrame in a single operation. This change has resulted in a performance increase of up to ~75%.