AttributeError: 'Float' object has no attribute 'sqrt'

Question

I'm trying to utilize the scipy.stats.linregress() function. I double checked that both my inputs are the same size and both are arrays. However I keep getting the following error:

Traceback (most recent call last):
  File "C:/Users/chianh/Desktop/Reliability Py Scripts/MedianRanks_Solver_Rev0.py", line 48, in <module>
    slope, intercept, r_value, p_value, std_err = stats.linregress(data, median_rank)
  File "C:\Users\chianh\AppData\Local\Programs\Python\Python35-32\lib\site-packages\scipy\stats\_stats_mstats_common.py", line 81, in linregress
    r_den = np.sqrt(ssxm * ssym)
AttributeError: 'Float' object has no attribute 'sqrt'

The code that i wrote is pasted below:

import numpy as np
import matplotlib.pyplot as plot
import scipy.stats as stats
from sympy import Symbol, binomial, expand_func, solve, solveset

def eval_median_rank( mylist ):
    P = 0.5                         # Percentage Point (would be 0.5 to be median)
    Z = Symbol('Z', real=True)      # Median Rank (Should be solved for each test unit)
    result = []                     # Place holder for the evaluated Median Ranks

    # Re-iterate and perform for each data point
    for j in range(1, mylist.size+1):
        eq = 0
        k = j

        # Generate the Median Rank equation which will be solved for 'Z' or Median Rank
        while k <= mylist.size:
            eq = eq + expand_func(binomial(mylist.size, k) * (Z**k) * (1 - Z)**(mylist.size-k))
            k += 1
        eq = eq - P

        # Solve for only Median Rank solutions that only fall between 0 and 1.
        # Then append to the returning list
        sol = [i for i in solve(eq, Z, list=True) if (i > 0) and (i < 1)]
        result.extend(sol)

    result.sort(reverse=True)
    return np.array(result)

######################################################################################################



# Note that the samples must be ordered in ascending fail-time/UOM,
# and the system must be solved in that order.

data = np.array([96, 257, 498, 763, 1051, 1744], dtype=object)
rank = np.arange(1, data.size+1)
median_rank = eval_median_rank(rank)
print(median_rank)

# Use Scipy's stats package to perform Least-Squares fit
slope, intercept, r_value, p_value, std_err = stats.linregress(data, median_rank)

Possible duplicate? stackoverflow.com/questions/18833639/… — Adib
– Adib, Commented Apr 13, 2016 at 20:59
Don't pass sympy objects to scipy functions. scipy doesn't handle sympy's Float object. — Warren Weckesser
– Warren Weckesser, Commented Apr 14, 2016 at 3:03

koraysal · Accepted Answer · 2025-04-15 13:35:09Z

-1

The error you're getting is caused because you're mixing SymPy objects with NumPy/SciPy, and SciPy's linregress() expects float or numpy.float64 types—not sympy.Float.

You're using dtype=object, and more importantly, in your eval_median_rank function, you're solving equations using SymPy, which returns results as sympy.Float, not Python float.

As a result, median_rank becomes a NumPy array of sympy.Float elements. When linregress() tries to compute things like np.sqrt(), it fails because sympy.Float doesn’t have .sqrt() (NumPy expects regular floats).

Solution:

Add this at the end of eval_median_rank():

return np.array(result, dtype=float)

This ensures the output is pure NumPy float64, fully compatible with SciPy.

answered Apr 15 at 13:35

koraysal

1

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

AttributeError: 'Float' object has no attribute 'sqrt'

1 Answer 1

Solution:

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Solution:

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related