I made this program here where you enter the feature values of many different samples of an object as “training data” for the program to recognize that object, the ratios of these features are then stored in an array and percentiles are calculated(using the numpy library) from this to create boundaries for the ratios.

The user then has to enter the feature values of an object being observed which gives the program the ratio of the features.
The program then compares this ratio to the percentiles or boundaries calculated previously using decision trees, this can hopefully give an estimate of the similarity between the object being observed and the object whose data was collected at the beginning of the program.

import numpy as np
between = False
exactly = False

#subject we want to classify
subject = input("enter the name of the test subject")

#number of training data entries:
num = int(input("how many data entries would you like to make?"))
#array to store training data
data = [[None]*2 for i in range(num)]
#ratios of training data:
multipliers = [[None] for i in range(num)]

#input for num number of training data values:
for i in range(0, num):
    #2 features for each data input:
    value = float(input("enter first value of subject " + str(i + 1)))
    valuetwo = float(input("enter secondvalue of subject " + str(i + 1)))
    #store feature values in 2d array:
    data[i][0] = value
    data[i][1] = valuetwo
    currentvalueone = data[i][0]
    currentvaluetwo = data[i][1]
    #this is the ratio between the two current features it describes the realtionship between the two features:
    currentmultiplier = currentvalueone / currentvaluetwo
    #store ratios in array
    multipliers[i] = currentmultiplier

#list ratios in order of size
multipliers.sort()

#array to store percentiles of ratios
percentiles = [None]*101


for i in range(101):
    percentiles[i] = np.percentile(multipliers, (i)*1)

#Put this part of the program in a function so that you can make as many comparisons as 
#you want: 
def enter():
    #enter feature values for unlabeled data
    first = float(input("enter first value"))
    second = float(input("enter second value"))
    #ratio for unlabeled data
    m = first / second    

    similarity = 0
    similarityl = 0
    similarityu = 0
    percentile = 0
    percentilel = 0
    percentileu = 0
    
    global percentiles
    
    #find correct percentile range:
    for i in range(101):
        #decision tree:
        if i != 101:
                if m >= percentiles[i] and m <= percentiles[i + 1] :
                    percentilel = i*1
                    percentileu = (i + 1)*1
                    if i*1 <= 50 and (i+1)*1 <= 50:
                        similarityl = 100 - (50 - ((i)*1))
                        similarityu = 100 - (50 - ((i+1)*1))
                    if i*1 >= 50 and (i + 1)*10 >= 50:
                        similarityl = 100 - (((i)*1) - 50)
                        similarityu = 100 - (((i + 1)*1) - 50)
                if m == percentiles[i]:
                    percentile = i
                    if i != 0 or i != 100:
                        if i*1 <= 50:
                            similarity = 100 - (50 - (i*1))
                        if i*1 >= 50:
                            similarity = 100 - ((i*1) - 50)  
                    if i == 0 or i == 100:
                        similarity = 100
                
              
    print(str((similarityl + similarityu)/2) + "% similarity to a " + subject)

    enter()

enter()

My question is, does this program do what i just said it does? Does it estimate the similarity between two objects based on input data?