Daily Average/Mean Using GroupBy

[ZandercraftGames]

I'm currently having an issue with getting a daily average grouped by date (a standardized string object - no need for datetime type here).

const dailyAvgDF = surveyFile.groupby(["DATE"]).col(["SPEED"]).mean();
console.log(dailyAvgDF);

This results in a dataframe of ["DATE", "SPEED_mean"] which is what I expect, but for some reason, every value is Infinity. I don't have any negative or NaN speeds. They are all integer values.

Here are my dtypes from surveyFile:

╔════════════╤══════════╗
║ DATE       │ string   ║
╟────────────┼──────────╢
║ TIME       │ string   ║
╟────────────┼──────────╢
║ SPEED      │ int32    ║
╟────────────┼──────────╢
║ DATETIME   │ string   ║
╟────────────┼──────────╢
║ DATETIME_A │ datetime ║
╚════════════╧══════════╝

It seems that the groupBy operation converts the numbers to a string type.

Seems related to #581 and #544

[ZandercraftGames]

I changed my code slightly to run this:

console.log(surveyFile.dtypes)
let grp = surveyFile.groupby(["DATE"])
console.log(grp.groups)
console.log(grp)
const avgs = grp.agg({SPEED: "mean", LENGTH: "sum"})
avgs.print()
console.log(avgs)
console.log(avgs.dtypes)

And it returns the following for sums:

"2.165.724.765.024.885.864.724.694.714.893.315.424.714.634.84.444.644.675.084.794.84.386.24.824 [...]"

Still Infinity for means.

Seems like it's trying to concat the strings or do addition operations on strings without maintaining the proper types.

[ZandercraftGames]

Some kinda sketchy workarounds, not sure what exactly did it, but this worked:

const copy = surveyFile.copy();
copy.addColumn(
  "SPEED",
  surveyFile["SPEED"].values.map((v: string) => {
    const num = Number(v);
    return isNaN(num) ? null : num;
  }),
  { inplace: true }
);
const formatted2 = copy.dropNa({ axis: 1, inplace: false })
let grp = formatted2.groupby(["DATE"])
const dailyAvgDF = grp.agg({SPEED: "mean"})
dailyAvgDF.print()

Still clearly something funky going on with the loss of type.