-2

I'm trying to find y = mx + b for a variety of different datasets. I've tried using:

slope_1, intercept_1 = linregress(values_1)

where values_1 is a Series type data.

bin_1 values
5th_per 10
25th_per 24
50th_per 28
75th_per 34
90th_per 50
95th_per 65

However, whenever I try to run the code, I get the error, IndexError: tuple index out of range.

I sort of understand the error, but am not sure how to fix this. Any suggestions on how to go about this or any other ways of finding the linear regression?

matrix_season
  • 159
  • 8
  • I'm assuming `values` are the y values, and the indices are the x values? – AJH Apr 08 '22 at 18:02
  • Yes, sorry, didn't make that clear initially. Is it possible to find a linear regression with string values? Or would I need to change those to actual number values in order to make it work? – matrix_season Apr 08 '22 at 18:07
  • Now I'm confused -- are you saying when values = 10, x = 5? And values = 24 corresponds to x = 25? Are your x values the row indices or the bin1 column? – AJH Apr 08 '22 at 18:10
  • Also, yes, any sort of regression is going to need numbers, not strings. – AJH Apr 08 '22 at 18:29

1 Answers1

2

Assuming values are the y values and the indices are the x values (e.g. values = 10 has x = 0, values = 24 has x = 1, etc.), then you can do:

values_1 = df.tolist("values")

# Convert values_1 to a numpy array and get x values as indices.
y = np.array(values_1, dtype=float)
x = np.arange(0, len(y), 1)

soln = np.polyfit(x, y, 1)
m, b = soln[0], soln[1]

Let me know if you have any questions.

EDIT: If you want to use the bin_1 values for the x values, replace the line x = np.arange in the code above with the following:

# Split each string according to _, creating a list for each string with the
# 1st element containing the number and the 2nd element being "per".
bins = df["bin_1"].str.split("_")

# Get the 1st element in each list using row[0].
# Then access just the number by discarding the ending "th" in each row[0].
bins = [row[0][:-2] for row in bins]
x = np.array(bins, dtype=float)
AJH
  • 799
  • 1
  • 3
  • 16