a-poor / a-poor/FeatureEng.jl

Create Transformation & Feature Eng Pipelines

Aberta
#1 1 comentário 0 reações 1 responsável Reivindicada por @a-poor Ver no GitHub
enhancement
Linguagem predominante
Julia
Estrelas
2
Forks
0
Métricas de merge de PRs
Nenhum PR com merge em 30d

Descrição

Using something like JSON to write code pipelines (as opposed to serializing functions -- since I don't think there's a good way to serialize Julia code in a way that will keep it consistent).

Doesn't necessarily have to be tied to Julia (could try implementing it in other languages, like Python). The pipeline could be planned out in python but carried out in Julia, for example.

## Example

Say you wanted to:
* Log scale column `"number-of-views"`
* Convert column `"timestamp"` to `DateTime` format
* Extract the day of the week from the column `"timestamp"`

The pipeline file might look like this:
```json
[
{
"operation":"log-scale",
"columns": [ "number-of-views" ],
"parameters": { "base": 10 }
},
{
"operation": "parse-datetime",
"columns": [ "timestamp" ],
"parameters":{
"inplace": true,
"format": "yyyy-mm-dd HH:MM:SS"
}
},
{
"operation": "extract-dow",
"columns": [ "timestamp" ],
"parameters": {
"new-column-name": "day-of-week"
}
}
]
```

Guia de contribuição

Nenhum guia de contribuição indexado para este repositório

Avaliação

Esta issue ainda não foi avaliada.

Receba novas issues na sua caixa de entrada

Um resumo curto de issues do GitHub para quem está começando.