Test_result_analysis

Sample calculation

The script for sample calculation use the material from Nist. This was used to evaluate if the sample size is suficient for taking conclusions about the data.

import math

z_1_a = 1.645 #z_{0.95} - significance level
z_1_b = 1.282 #z_{0.90} - risk willing to take

p0=0.1 #p denoting the proportion of defectives already assumed

delta = 0.10 #change in the proportion defective that we are interested in detecting 

p1 = 0.20 # delta = abs(p1-p0)

N = ((z_1_a*math.sqrt(p0*(1-p0))+z_1_b*math.sqrt(p1*(1-p1)))/delta)**2

print(round(N,3))

For this analysis, the ideal sample size would be 102.

Data separation

Reading the csv file, the data was separated and stored in diferent vectors, which one for a test condition.

    cond_1=[]
    for i in range(1, len(data)):
            if len(data[i]) >= 5 and int(data[i][1])==433 and int(data[i][2])==14 and int(data[i][3])==7 and int(data[i][4])==0:
                cond_1.append(data[i])
    cond_2=[]
    for i in range(1, len(data)):
            if len(data[i]) >= 5 and int(data[i][1])==433 and int(data[i][2])==14 and int(data[i][3])==7 and int(data[i][4])==2:
                cond_2.append(data[i])

Parameters Calculation

In this script, the mean, variance, std deviation and confidence interval were calculated.

for condition_num in range(1, num_conditions + 1):
    # Assuming TX_data.cond_1, TX_data.cond_2, ..., TX_data.cond_40 exist
    condition_data = getattr(TX_LoRa_data_filtering, f"cond_{condition_num}")

    # Extract power values for the current condition
    power_values = [float(row[5]) for row in condition_data[1:]]

    # Calculate parameters for the current condition
    x_barra = round(np.mean(power_values), 4)
    variancia = round(np.var(power_values), 4)
    std_dev = round(np.std(power_values), 4)

    # Calculate t-statistic and confidence interval
    n = len(power_values)
    gl = n - 1
    # Nível de confiança
    p = 0.90
    # Complementar
    alpha = 1 - p
    ts = stats.t.ppf(alpha, gl)
    tol = round(ts * std_dev / (math.sqrt(n)), 4)

The script also stores the mean and the confidence intervals for each condition, so we can analyse the samples condensed in each condition.