Fix - Frequencies should be Double

[kayvank]

Frequencies should show doubles, not Int, per #42

[kayvank]

I am still stuck on this issue, and would very much like to learn how to fix it.
I thought maybe doing this would help:

    in L.foldl' (\df (col, k) -> insertVector (vText col) (V.fromList [toRowValue k, toRowValue $ k * 100.0  / total]) df) initDf counts

it did not and resulted compile error:

Could not deduce ‘Read DataFrame.Internal.Row.RowValue’
        arising from a use of ‘insertVector’
      from the context: Columnable a
        bound by a pattern with constructor:
                   BoxedColumn :: forall a. Columnable a => V.Vector a -> Column,
                 in a case alternative

[mchav]

I see. I don't quite like my approach here but for something to be Columnable it has to have a Read, Ord, Eq, and Show instance.

ghci> data Test = T Int deriving (Show, Eq, Read, Ord)
ghci> D.fromList [T 1, T 2]
[T 1,T 2]

RowValue doesn't have a Read instance and you can't do deriving similar to what we did in the CLI cause it's a GADT.

So to unblock you can just make a stub Read instance for RowValue which throws UNIMPLEMENTED. I don't imagine someone will try to read row values in the near future. I don't even know where they are used.

Sorry this was so painful. It ended up suffering from silly design decisions.

[ghost]

You can do a quick and dirty Read instance implementation:

instance Read RowValue where
  readsPrec prec ('V': 'a': 'l': 'u': 'e': ' ' : a) = readsPrec prec a -- or perhaps Value $ readsPrec prec a ?

At least, that's the legacy way of doing it. The newer better way is probably to do readPrec (instead of readsPrec) and use lexP or something

[mchav]

Can we try leaving this as an int instead?

[mchav]

total can also be an int

[mchav]

For percentage we can do the fromIntegral just before division rather than having the counts be doubles.

We can also format the string to round it for display:

import Text.Printf (printf)

printf "%.2f\n" (3.14159 :: Double)

[kayvank]

The printf will result percentage to print as such:

 Percentage (%) | "44.26"   | "31.74"  | "0.02"   | "11.09"  | "12.88"

I could read them back as double priort to converting to RowValue which will result in this:

1     | Percentage (%) | 44.26     | 31.74    | 2.0e-2   | 11.09    | 12.88

I'll lookinto rounding the Doubles as an alterntive

[kayvank]

I used the round function that you already had there and here is the result:

--------------------------------------------------------------------------------
index |   Statistic    | <1H OCEAN |  INLAND  |  ISLAND  | NEAR BAY | NEAR OCEAN
------|----------------|-----------|----------|----------|----------|-----------
 Int  |      Text      | RowValue  | RowValue | RowValue | RowValue |  RowValue
------|----------------|-----------|----------|----------|----------|-----------
0     | Count          | 9136      | 6551     | 5        | 2290     | 2658
1     | Percentage (%) | 44.26     | 31.74    | 2.0e-2   | 11.09    | 12.8

[mchav]

This looks good! Thank you.