Timeseries GUI: button generates Type Error on comparision to a dataframe column

Question

Objective:

I am writing an application (using Tkinter, Python 3.0 - ubuntu). The event trigger is user input a float, user presses a button; then 'calculate' function compares a column from a pandas dataframe (made from a .csv) to the user's input.

It should return the five rows.

Those rows are determined by the proximity (closeness) of the user's number compared with all floats in a particular column of the dataframe the function created using pandas. Next, it should return summary stats for (one 0f the columns) from those 5 returned rows. (Note: pandas' df.column.describe() will suffice for now).

(I have a pressing deadline, so any thoughtful suggestions will be met with immediate good karma and instant upvoting). :-)

Error1:

   **TypeError: unsupported operand type(s) for -: 'float' and 'instance'**

Note1:

I know they are missed typed -- are there convenient hacks around this?

The compiler considers the floats from dataframe and 'instance' my declared es_in=DoubleVar().

Modification: Tkinter has no 'FloatVar()'. I cast float(es_in). It gives:

Error2:

  AttributeError: DoubleVar instance has no attribute '__float__'




import pandas as pd
from Tkinter import *
import ttk

def calculate(*args):
    try:
        df=pd.read_csv('master_pl.csv')

Note2:

The following line gives me the five 'closest' rows. The error is generated from this line: where I compare a column of my dataframe: df["ES"], to the user-defined input called es_in.

        df.ix[(df["ES"][:]-es_in).abs().argsort()[:5]]

     except ValueError:
       pass


  <___calculate_function_ends_here___>
   # GUI code continues...


root = Tk()
root.title("Cluster Program")

mainframe = ttk.Frame(root, padding="3 3 12 12")
mainframe.grid(column=0, row=0, sticky=(N, W, E, S))
mainframe.columnconfigure(0, weight=1)
mainframe.rowconfigure(0, weight=1)

The user defined input is here:

es_in = DoubleVar()

es_entry = ttk.Entry(mainframe, width=7)
es_entry.grid(column=2, row=1, sticky=(W, E))


ttk.Button(mainframe, text="Calculate", command=calculate).grid(column=3, row=3, sticky=W)

 ttk.Label(mainframe, text="ES").grid(column=3, row=1, sticky=W)
 ttk.Label(mainframe, text="output_fiveclosestrows").grid(column=3, row=2, sticky=W)



es_entry.focus()
root.bind('<Return>', calculate) 


<end>

Answer 1

If I understand you correctly, the error is produced by this bit of code: df["ES"][:]-es_in, correct?

es_in is an instance of DoubleVar, which means you can't use the value of that variable directly since the value is an instance rather than a value -- exactly what the error message is telling you.

So, how do you get the value from a DoubleVar? With the get method. Change the code to this, and see if it works any better:

df.ix[(df["ES"][:]-es_in.get()).abs().argsort()[:5]]

Notice where it uses es_in.get() rather than es_in

Of course, another problem seems to be that you're subtracting a floating point value from a sequence, which makes no sense. Are you wanting to subtract it from each value in the sequence?

It would help if you broke your code up into multiple lines. It would make the code easier to understand, and easier for you to debug. For example, if I break your code into multiple lines, I end up with something like this:

diff = df["ES"][:]-es_in.get()
delta = diff.abs()
sorted = delta.argsort()
index = sorted[:5]
df.ix[index]

That looks wrong to me, but I don't know numpy or pandas, so maybe it makes sense in that context. Regardless, breaking a complex statement into multiple statements is a good first step in trying to debug code that isn't working the way you behave. After breaking it apart you'll have a better idea of what part of the code is failing, and you can use a debugger or print statements to display intermediate values to validate your assumptions.