๋ณธ๋ฌธ ๋ฐ”๋กœ๊ฐ€๊ธฐ

728x90

metrics

4/5 ํ™” ํ™”์š”์ผ! Logistic Regression์„ ํ™œ์šฉํ•ด ๋จธ์‹ ๋Ÿฌ๋‹ ์ง„ํ–‰ ์‹œ ์ฃผ์˜์‚ฌํ•ญ์„ ์•Œ์•„๋ณธ๋‹ค. ์•ž์œผ๋กœ ์šฐ๋ฆฌ๋Š” Classification(์ดํ•ญ๋ถ„๋ฅ˜)์˜ Metrics๋กœ Accuracy๋ฅผ ์‚ฌ์šฉํ•  ์˜ˆ์ •์ด๋‹ค. ๋ชจ๋ธ ํ‰๊ฐ€ ์ „ ๊ณ ๋ คํ•ด์•ผ ํ•˜๋Š” ๊ฒƒ๋“ค 1. learning rate(ํ•™์Šต๋ฅ ) : loss ๊ฐ’์„ ๋ณด๋ฉด์„œ ํ•™์Šต๋ฅ ์„ ์กฐ์ •ํ•ด์•ผ ํ•จ. ๋ณดํ†ต 1์˜ ๋งˆ์ด๋„ˆ์Šค 4์Šน์œผ๋กœ ์žก์Œ ํ•™์Šต๋ฅ ์ด ๋„ˆ๋ฌด ํฌ๋‹ค๋ฉด global minima(W')๋ฅผ ์ฐพ์„ ์ˆ˜ ์—†๊ฒŒ ๋จ โ†’ OverShooting ๋ฐœ์ƒ ํ•™์Šต๋ฅ ์ด ์•„์ฃผ ์ž‘๋‹ค๋ฉด local minima ์ฐพ๊ฒŒ ๋จ 2. Normalization(์ •๊ทœํ™”) : MinMax Scaling - 0 ~ 1. ์ด์ƒ์น˜์— ๋ฏผ๊ฐํ•จ Standardization - ํ‘œ์ค€ํ™”, Z-Score. ์ƒ๋Œ€์ ์œผ๋กœ ์ด์ƒ์น˜์— ๋‘”๊ฐํ•จ, ๋ชจ๋“  ์นผ๋Ÿผ์—.. ๋”๋ณด๊ธฐ
4/4 ์›” ์›”์š”์ผ! ์˜ค๋Š˜์€ ๊ธˆ์š”์ผ์— ์‹ค์Šต ์˜ˆ์ œ๋กœ ์ฃผ์–ด์กŒ๋˜ admission(๋Œ€ํ•™์› ํ•ฉ๊ฒฉ ์—ฌ๋ถ€) ๋ฐ์ดํ„ฐ์…‹์„ Sklearn, Tensorflow๋กœ ๊ตฌํ˜„ํ•˜๊ณ , ์ง€๋‚œ์ฃผ์— ๋ฐฐ์šด Logistic Regression์„ ํ™œ์šฉํ•ด ํ‰๊ฐ€์ง€ํ‘œ(Metrics)๋ฅผ ์•Œ์•„๋ณธ๋‹ค. 1. Logistic Regression by Sklearn import numpy as np import pandas as pd import tensorflow as tf from sklearn import linear_model from sklearn.preprocessing import MinMaxScaler from scipy import stats import matplotlib.pyplot as plt import warnings warnings.filter.. ๋”๋ณด๊ธฐ

728x90