由于是刚刚涉及数据挖掘,所以个人理解部分会有很多纰漏,望指正
Tip:本次新人赛是Datawhale与天池联合发起的零基础入门系列赛事第五场 —— 零基础入门心电图心跳信号多分类预测挑战赛。
2016年6月,国务院办公厅印发《国务院办公厅关于促进和规范健康医疗大数据应用发展的指导意见》,文件指出健康医疗大数据应用发展将带来健康医疗模式的深刻变化,有利于提升健康医疗服务效率和质量。
赛题以心电图数据为背景,要求选手根据心电图感应数据预测心跳信号,其中心跳信号对应正常病例以及受不同心律不齐和心肌梗塞影响的病例,这是一个多分类的问题。通过这道赛题来引导大家了解医疗大数据的应用,帮助竞赛新人进行自我练习、自我提高。
比赛地址:https://tianchi.aliyun.com/competition/entrance/531883/introduction
比赛要求参赛选手根据给定的数据集,建立模型,预测不同的心跳信号。赛题以预测心电图心跳信号类别为任务,数据集报名后可见并可下载,该该数据来自某平台心电图数据记录,总数据量超过20万,主要为1列心跳信号序列数据,其中每个样本的信号序列采样频次一致,长度相等。为了保证比赛的公平性,将会从中抽取10万条作为训练集,2万条作为测试集A,2万条作为测试集B,同时会对心跳信号类别(label)信息进行脱敏。
通过这道赛题来引导大家走进医疗大数据的世界,主要针对于于竞赛新人进行自我练习,自我提高。
一般而言,对于数据在比赛界面都有对应的数据概况介绍(匿名特征除外),说明列的性质特征。了解列的性质会有助于我们对于数据的理解和后续分析。
Tip:匿名特征,就是未告知数据列所属的性质的特征列。
train.csv
testA.csv
选手需提交4种不同心跳信号预测的概率,选手提交结果与实际心跳类型结果进行对比,求预测的概率与真实值差值的绝对值。
具体计算公式如下:
总共有n个病例,针对某一个信号,若真实值为[y1,y2,y3,y4],模型预测概率值为[a1,a2,a3,a4],那么该模型的评价指标abs-sum为
a b s − s u m = ∑ j = 1 n ∑ i = 1 4 ∣ y i − a i ∣ {abs-sum={\mathop{ \sum }\limits_{{j=1}}^{{n}}{{\mathop{ \sum }\limits_{{i=1}}^{{4}}{{ \left| {y\mathop{{}}\nolimits_{{i}}-a\mathop{{}}\nolimits_{{i}}} \right| }}}}}} abs−sum=j=1∑ni=1∑4∣yi−ai∣
例如,某心跳信号类别为1,通过编码转成[0,1,0,0],预测不同心跳信号概率为[0.1,0.7,0.1,0.1],那么这个信号预测结果的abs-sum为
a b s − s u m = ∣ 0.1 − 0 ∣ + ∣ 0.7 − 1 ∣ + ∣ 0.1 − 0 ∣ + ∣ 0.1 − 0 ∣ = 0.6 {abs-sum={ \left| {0.1-0} \right| }+{ \left| {0.7-1} \right| }+{ \left| {0.1-0} \right| }+{ \left| {0.1-0} \right| }=0.6} abs−sum=∣0.1−0∣+∣0.7−1∣+∣0.1−0∣+∣0.1−0∣=0.6
多分类算法常见的评估指标如下:
其实多分类的评价指标的计算方式与二分类完全一样,只不过我们计算的是针对于每一类来说的召回率、精确度、准确率和 F1分数。
1、混淆矩阵(Confuse Matrix)
第一个字母T/F,表示预测的正确与否;第二个字母P/N,表示预测的结果为正例或者负例。如TP就表示预测对了,预测的结果是正例,那它的意思就是把正例预测为了正例。
2.准确率(Accuracy)
准确率是常用的一个评价指标,但是不适合样本不均衡的情况,医疗数据大部分都是样本不均衡数据。
A c c u r a c y = C o r r e c t T o t a l A c c u r a c y = T P + T N T P + T N + F P + F N Accuracy=\frac{Correct}{Total}\\ Accuracy = \frac{TP + TN}{TP + TN + FP + FN} Accuracy=TotalCorrectAccuracy=TP+TN+FP+FNTP+TN
3、精确率(Precision)也叫查准率简写为P
精确率(Precision)是针对预测结果而言的,其含义是在被所有预测为正的样本中实际为正样本的概率在被所有预测为正的样本中实际为正样本的概率,精确率和准确率看上去有些类似,但是是两个完全不同的概念。精确率代表对正样本结果中的预测准确程度,准确率则代表整体的预测准确程度,包括正样本和负样本。
P r e c i s i o n = T P T P + F P Precision = \frac{TP}{TP + FP} Precision=TP+FPTP
4.召回率(Recall) 也叫查全率 简写为R
召回率(Recall)是针对原样本而言的,其含义是在实际为正的样本中被预测为正样本的概率。
R e c a l l = T P T P + F N Recall = \frac{TP}{TP + FN} Recall=TP+FNTP
下面我们通过一个简单例子来看看精确率和召回率。假设一共有10篇文章,里面4篇是你要找的。根据你的算法模型,你找到了5篇,但实际上在这5篇之中,只有3篇是你真正要找的。
那么算法的精确率是3/5=60%,也就是你找的这5篇,有3篇是真正对的。算法的召回率是3/4=75%,也就是需要找的4篇文章,你找到了其中三篇。以精确率还是以召回率作为评价指标,需要根据具体问题而定。
5.宏查准率(macro-P)
计算每个样本的精确率然后求平均值
m a c r o P = 1 n ∑ 1 n p i {macroP=\frac{{1}}{{n}}{\mathop{ \sum }\limits_{{1}}^{{n}}{p\mathop{{}}\nolimits_{{i}}}}} macroP=n11∑npi
6.宏查全率(macro-R)
计算每个样本的召回率然后求平均值
m a c r o R = 1 n ∑ 1 n R i {macroR=\frac{{1}}{{n}}{\mathop{ \sum }\limits_{{1}}^{{n}}{R\mathop{{}}\nolimits_{{i}}}}} macroR=n11∑nRi
7.宏F1(macro-F1)
m a c r o F 1 = 2 × m a c r o P × m a c r o R m a c r o P + m a c r o R {macroF1=\frac{{2 \times macroP \times macroR}}{{macroP+macroR}}} macroF1=macroP+macroR2×macroP×macroR
与上面的宏不同,微查准查全,先将多个混淆矩阵的TP,FP,TN,FN对应位置求平均,然后按照P和R的公式求得micro-P和micro-R,最后根据micro-P和micro-R求得micro-F1
8.微查准率(micro-P)
m i c r o P = T P ‾ T P ‾ × F P ‾ {microP=\frac{{\overline{TP}}}{{\overline{TP} \times \overline{FP}}}} microP=TP×FPTP
9.微查全率(micro-R)
m i c r o R = T P ‾ T P ‾ × F N ‾ {microR=\frac{{\overline{TP}}}{{\overline{TP} \times \overline{FN}}}} microR=TP×FNTP
10.微F1(micro-F1)
m i c r o F 1 = 2 × m i c r o P × m i c r o R m i c r o P + m i c r o R {microF1=\frac{{2 \times microP\times microR }}{{microP+microR}}} microF1=microP+microR2×microP×microR
报名成功后,选手下载数据,在本地调试算法,每天可提交3次结果;
提交后将进行实时评测;每天排行榜更新时间为12:00和20:00,按照评测指标得分从高到低排序;排行榜将选择历史最优成绩进行展示;
import os
import gc
import math
import pandas as pd
import numpy as np
import lightgbm as lgb
import xgboost as xgb
from catboost import CatBoostRegressor
from sklearn.linear_model import SGDRegressor, LinearRegression, Ridge
from sklearn.preprocessing import MinMaxScaler
from sklearn.model_selection import StratifiedKFold, KFold
from sklearn.metrics import log_loss
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import OneHotEncoder
from tqdm import tqdm
import matplotlib.pyplot as plt
import time
import warnings
warnings.filterwarnings('ignore')
train = pd.read_csv('train.csv')
test=pd.read_csv('testA.csv')
train.head()
id | heartbeat_signals | label |
---|---|---|
0 | 0.9912297987616655,0.9435330436439665,0.764677… | 0.0 |
1 | 0.9912297987616655,0.9435330436439665,0.764677… | 0.0 |
2 | 1.0,0.9591487564065292,0.7013782792997189,0.23… | 2.0 |
3 | 0.9757952826275774,0.9340884687738161,0.659636… | 0.0 |
4 | 0.0,0.055816398940721094,0.26129357194994196,0… | 2.0 |
test.head()
id | hearbeat_signals |
---|---|
100000 | 0.9915713654170097,1.0,0.6318163407681274,0.13… |
100001 | 0.6075533139615096,0.5417083883163654,0.340694… |
100002 | 0.9752726292239277,0.6710965234906665,0.686758… |
100003 | 0.9956348033996116,0.9170249621481004,0.521096… |
100004 | 1.0,0.8879490481178918,0.745564725322326,0.531… |
def reduce_mem_usage(df):
start_mem = df.memory_usage().sum() / 1024**2
print('Memory usage of dataframe is {:.2f} MB'.format(start_mem))
for col in df.columns:
col_type = df[col].dtype
if col_type != object:
c_min = df[col].min()
c_max = df[col].max()
if str(col_type)[:3] == 'int':
if c_min > np.iinfo(np.int8).min and c_max < np.iinfo(np.int8).max:
df[col] = df[col].astype(np.int8)
elif c_min > np.iinfo(np.int16).min and c_max < np.iinfo(np.int16).max:
df[col] = df[col].astype(np.int16)
elif c_min > np.iinfo(np.int32).min and c_max < np.iinfo(np.int32).max:
df[col] = df[col].astype(np.int32)
elif c_min > np.iinfo(np.int64).min and c_max < np.iinfo(np.int64).max:
df[col] = df[col].astype(np.int64)
else:
if c_min > np.finfo(np.float16).min and c_max < np.finfo(np.float16).max:
df[col] = df[col].astype(np.float16)
elif c_min > np.finfo(np.float32).min and c_max < np.finfo(np.float32).max:
df[col] = df[col].astype(np.float32)
else:
df[col] = df[col].astype(np.float64)
else:
df[col] = df[col].astype('category')
end_mem = df.memory_usage().sum() / 1024**2
print('Memory usage after optimization is: {:.2f} MB'.format(end_mem))
print('Decreased by {:.1f}%'.format(100 * (start_mem - end_mem) / start_mem))
return df
# 简单预处理
train_list = []
for items in train.values:
train_list.append([items[0]] + [float(i) for i in items[1].split(',')] + [items[2]])
train = pd.DataFrame(np.array(train_list))
train.columns = ['id'] + ['s_'+str(i) for i in range(len(train_list[0])-2)] + ['label']
train = reduce_mem_usage(train)
test_list=[]
for items in test.values:
test_list.append([items[0]] + [float(i) for i in items[1].split(',')])
test = pd.DataFrame(np.array(test_list))
test.columns = ['id'] + ['s_'+str(i) for i in range(len(test_list[0])-1)]
test = reduce_mem_usage(test)
Memory usage of dataframe is 157.93 MB
Memory usage after optimization is: 39.67 MB
Decreased by 74.9%
Memory usage of dataframe is 31.43 MB
Memory usage after optimization is: 7.90 MB
Decreased by 74.9%
x_train = train.drop(['id','label'], axis=1)
y_train = train['label']
x_test=test.drop(['id'], axis=1)
def abs_sum(y_pre,y_tru):
y_pre=np.array(y_pre)
y_tru=np.array(y_tru)
loss=sum(sum(abs(y_pre-y_tru)))
return loss
def cv_model(clf, train_x, train_y, test_x, clf_name):
folds = 5
seed = 2021
kf = KFold(n_splits=folds, shuffle=True, random_state=seed)
test = np.zeros((test_x.shape[0],4))
cv_scores = []
onehot_encoder = OneHotEncoder(sparse=False)
for i, (train_index, valid_index) in enumerate(kf.split(train_x, train_y)):
print('************************************ {} ************************************'.format(str(i+1)))
trn_x, trn_y, val_x, val_y = train_x.iloc[train_index], train_y[train_index], train_x.iloc[valid_index], train_y[valid_index]
if clf_name == "lgb":
train_matrix = clf.Dataset(trn_x, label=trn_y)
valid_matrix = clf.Dataset(val_x, label=val_y)
params = {
'boosting_type': 'gbdt',
'objective': 'multiclass',
'num_class': 4,
'num_leaves': 2 ** 5,
'feature_fraction': 0.8,
'bagging_fraction': 0.8,
'bagging_freq': 4,
'learning_rate': 0.1,
'seed': seed,
'nthread': 28,
'n_jobs':24,
'verbose': -1,
}
model = clf.train(params,
train_set=train_matrix,
valid_sets=valid_matrix,
num_boost_round=2000,
verbose_eval=100,
early_stopping_rounds=200)
val_pred = model.predict(val_x, num_iteration=model.best_iteration)
test_pred = model.predict(test_x, num_iteration=model.best_iteration)
val_y=np.array(val_y).reshape(-1, 1)
val_y = onehot_encoder.fit_transform(val_y)
print('预测的概率矩阵为:')
print(test_pred)
test += test_pred
score=abs_sum(val_y, val_pred)
cv_scores.append(score)
print(cv_scores)
print("%s_scotrainre_list:" % clf_name, cv_scores)
print("%s_score_mean:" % clf_name, np.mean(cv_scores))
print("%s_score_std:" % clf_name, np.std(cv_scores))
test=test/kf.n_splits
return test
def lgb_model(x_train, y_train, x_test):
lgb_test = cv_model(lgb, x_train, y_train, x_test, "lgb")
return lgb_test
lgb_test = lgb_model(x_train, y_train, x_test)
************************************ 1 ************************************
[LightGBM] [Warning] num_threads is set with nthread=28, will be overridden by n_jobs=24. Current value: num_threads=24
Training until validation scores don't improve for 200 rounds
[100] valid_0's multi_logloss: 0.0525735
[200] valid_0's multi_logloss: 0.0422444
[300] valid_0's multi_logloss: 0.0407076
[400] valid_0's multi_logloss: 0.0420398
Early stopping, best iteration is:
[289] valid_0's multi_logloss: 0.0405457
预测的概率矩阵为:
[[9.99969791e-01 2.85197261e-05 1.00341946e-06 6.85357631e-07]
[7.93287264e-05 7.69060914e-04 9.99151590e-01 2.00810971e-08]
[5.75356884e-07 5.04051497e-08 3.15322414e-07 9.99999059e-01]
...
[6.79267940e-02 4.30206297e-04 9.31640185e-01 2.81516302e-06]
[9.99960477e-01 3.94098074e-05 8.34030725e-08 2.94638661e-08]
[9.88705846e-01 2.14081630e-03 6.67418381e-03 2.47915423e-03]]
[607.0736049372186]
************************************ 2 ************************************
[LightGBM] [Warning] num_threads is set with nthread=28, will be overridden by n_jobs=24. Current value: num_threads=24
Training until validation scores don't improve for 200 rounds
[100] valid_0's multi_logloss: 0.0566626
[200] valid_0's multi_logloss: 0.0450852
[300] valid_0's multi_logloss: 0.044078
[400] valid_0's multi_logloss: 0.0455546
Early stopping, best iteration is:
[275] valid_0's multi_logloss: 0.0437793
预测的概率矩阵为:
[[9.99991401e-01 7.69109547e-06 6.65504756e-07 2.42084688e-07]
[5.72380482e-05 1.32812809e-03 9.98614607e-01 2.66534396e-08]
[2.82123411e-06 4.13195205e-07 1.34026965e-06 9.99995425e-01]
...
[6.96398024e-02 6.52459907e-04 9.29685742e-01 2.19960932e-05]
[9.99972366e-01 2.75069005e-05 7.68142933e-08 5.07415018e-08]
[9.67263676e-01 7.26154408e-03 2.41533542e-02 1.32142531e-03]]
[607.0736049372186, 623.4313863731124]
************************************ 3 ************************************
[LightGBM] [Warning] num_threads is set with nthread=28, will be overridden by n_jobs=24. Current value: num_threads=24
Training until validation scores don't improve for 200 rounds
[100] valid_0's multi_logloss: 0.0498722
[200] valid_0's multi_logloss: 0.038028
[300] valid_0's multi_logloss: 0.0358066
[400] valid_0's multi_logloss: 0.0361478
[500] valid_0's multi_logloss: 0.0379597
Early stopping, best iteration is:
[340] valid_0's multi_logloss: 0.0354344
预测的概率矩阵为:
[[9.99972032e-01 2.62406774e-05 1.17282152e-06 5.54230651e-07]
[1.05242811e-05 6.50215805e-05 9.99924453e-01 6.93812546e-10]
[1.93240868e-06 1.10384984e-07 3.76773426e-07 9.99997580e-01]
...
[1.34894410e-02 3.84569683e-05 9.86471555e-01 5.46564350e-07]
[9.99987431e-01 1.25532882e-05 1.03902298e-08 5.46727770e-09]
[9.78722948e-01 1.06329839e-02 6.94192038e-03 3.70214810e-03]]
[607.0736049372186, 623.4313863731124, 508.02381607269535]
************************************ 4 ************************************
[LightGBM] [Warning] num_threads is set with nthread=28, will be overridden by n_jobs=24. Current value: num_threads=24
Training until validation scores don't improve for 200 rounds
[100] valid_0's multi_logloss: 0.0564768
[200] valid_0's multi_logloss: 0.0448698
[300] valid_0's multi_logloss: 0.0446719
[400] valid_0's multi_logloss: 0.0470399
Early stopping, best iteration is:
[250] valid_0's multi_logloss: 0.0438853
预测的概率矩阵为:
[[9.99979692e-01 1.70821979e-05 1.27048476e-06 1.95571841e-06]
[5.66207785e-05 4.02275314e-04 9.99541086e-01 1.82828519e-08]
[2.62267451e-06 3.58613522e-07 4.78645006e-06 9.99992232e-01]
...
[4.56636552e-02 5.69497433e-04 9.53758468e-01 8.37980573e-06]
[9.99896785e-01 1.02796802e-04 2.46636563e-07 1.72061021e-07]
[8.70911669e-01 1.73790185e-02 1.04478175e-01 7.23113697e-03]]
[607.0736049372186, 623.4313863731124, 508.02381607269535, 660.4867407547267]
************************************ 5 ************************************
[LightGBM] [Warning] num_threads is set with nthread=28, will be overridden by n_jobs=24. Current value: num_threads=24
Training until validation scores don't improve for 200 rounds
[100] valid_0's multi_logloss: 0.0506398
[200] valid_0's multi_logloss: 0.0396422
[300] valid_0's multi_logloss: 0.0381065
[400] valid_0's multi_logloss: 0.0390162
[500] valid_0's multi_logloss: 0.0414986
Early stopping, best iteration is:
[324] valid_0's multi_logloss: 0.0379497
预测的概率矩阵为:
[[9.99993352e-01 6.02902202e-06 1.13002685e-07 5.06277302e-07]
[1.03959552e-05 5.03778956e-04 9.99485820e-01 5.07638601e-09]
[1.92568065e-07 5.07155306e-08 4.94690856e-08 9.99999707e-01]
...
[8.83103121e-03 2.51969353e-05 9.91142776e-01 9.96143937e-07]
[9.99984791e-01 1.51997858e-05 5.62426491e-09 3.80450197e-09]
[9.86084001e-01 8.75968498e-04 1.09742304e-02 2.06580027e-03]]
[607.0736049372186, 623.4313863731124, 508.02381607269535, 660.4867407547267, 539.2160054696063]
lgb_scotrainre_list: [607.0736049372186, 623.4313863731124, 508.02381607269535, 660.4867407547267, 539.2160054696063]
lgb_score_mean: 587.646310721472
lgb_score_std: 55.94453640571462
temp=pd.DataFrame(lgb_test)
result=pd.read_csv('sample_submit.csv')
result['label_0']=temp[0]
result['label_1']=temp[1]
result['label_2']=temp[2]
result['label_3']=temp[3]
result.to_csv('submit.csv',index=False)
1.根据现有baseline继续学习,在特征工程部分人为制造特征,然后排除无用特征;
2.ML训练模型使用lightgbm和caboost为主,在连个树模型都训练好后做模型融合,用二层贝叶斯回归尝试降低分数;
3.DL训练模型首先使用NN模型尝试,如果效果比较好则尝试更多深度学习算法。
4.其他优化思路待补充…