该代码实现了一个使用Streamlit的交互式应用，用户可以选择文件类型，预览数据，进行数据标准化，选择聚类算法和数量，展示雷达图，并进行分类模型训练，包括决策树、KNN和朴素贝叶斯。应用支持通过局域网共享，便于协作和演示。
这里的前置工作是数据分析，将特征值抽取出来
def ALG(ALG):
if ALG == "机器学习模型训练":
st.write(ALG)
read = st.sidebar.selectbox("读取文件类型", ("csv", "excle"))
title = st.sidebar.text_input('输入地址', 'E:/桌面/项目实训/电商优惠券特征数据')
flie = pd.read_csv(title+"."+read, encoding="GBK")
n = st.select_slider("滑动显示数量", options=[5, 10, 20, 50, 100])
show = st.selectbox("是否显示表", ("否", "是"))
if show == "否":
st.write("不显示")
elif show == "是":
st.dataframe(flie[0:n])
else:
pass
Fl = st.multiselect("选择特征列",flie.columns)
if Fl !=[]:
choose = flie[Fl]
st.write(choose.describe())
from sklearn.preprocessing import StandardScaler, MinMaxScaler
md = st.selectbox("数据标准化", ("标准差标准化StandardScaler"
, "最大最小标准化MinMaxScaler"))
if md == "标准差标准化StandardScaler":
scale = StandardScaler()
elif md == "最大最小标准化MinMaxScaler":
scale = MinMaxScaler()
else:
pass
cf = st.selectbox("标准化", ("是", "否"))
if cf =="是":
feature = scale.fit_transform(choose)
st.write("均值：",feature.mean())
st.write("方差：",feature.std())
from sklearn.cluster import KMeans
num_c = list(range(2, 12))
num_c1 = st.selectbox("选择聚类数量", num_c)
RS = st.text_input("聚类随机种子默认0",0)
cf = st.selectbox("选择聚类",("否","是"))
if cf =="是":
model = KMeans(n_clusters=int(num_c1), random_state=int(RS))
model.fit(feature)
st.write("聚类中心",model.cluster_centers_)
labels = st.select_slider("显示数量", options=[20, 50, 100])
st.write("聚类标签",model.labels_[0:labels])
st.write("聚类标签与数量",pd.Series(model.labels_).value_counts())
def radarplot(model_center=None, label=None):
import matplotlib.pyplot as plt
plt.rcParams['font.sans-serif'] = ['SimHei'] # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False # 用来正常显示负号
n = len(label) # 特征个数
# 对labels进行封闭，否则会有因为matplotlib版本引起的错误
label = np.concatenate((label, [label[0]]))
# 间隔采样，设置雷达图的角度，用于平分切开一个圆面,endpoint设置为False表示随机采样不包括stop的值
angles = np.linspace(0, 2 * np.pi, n, endpoint=False)
# 拼接多个数组，使雷达图一圈封闭起来
angles = np.concatenate((angles, [angles[0]]))
# 创建一个空白画布
fig = plt.figure(figsize=(8, 8))
fig, ax = plt.subplots()#这个语句是用来传递图形的
# 创建子图，设置极坐标格式，绘制圆形
ax = fig.add_subplot(1, 1, 1, polar=True)
# 添加每个特征的标签
ax.set_thetagrids(angles * 180 / np.pi, label)
# 设置y轴范围
ax.set_ylim(model_center.min(), model_center.max())
# 添加网格线
ax.grid(True)
# 设置备选的折线颜色和样式,防止线条重复
sam = ['r', 'pink', 'g', 'b', 'm', 'y', 'k', '0', 'c']
mak = ['4', '8', 'x', '*', 'd', '_', '.', '+', '|']
labels = []
# 循环添加每个类别的线圈
for i in range(len(model_center)):
values = np.concatenate((model_center[i], [model_center[i][0]]))
ax.plot(angles, values, c=sam[i], marker=mak[i])
plt.yticks(fontsize=15)
plt.xticks(fontsize=15)
labels.append(str(i)+"类")
plt.legend(labels, fontsize=15)
st.pyplot(fig)
show_radarplot = st.selectbox("是否显示雷达图", ("是", "否"))
if show_radarplot =="是":
radarplot(model.cluster_centers_, choose.columns)
else:
pass
num_s = list([0.1,0.15,0.2, 0.25,0.3,0.35, 0.4,0.45, 0.5, 0.55,0.6,0.65, 0.7,0.75, 0.8,0.85, 0.9])
num_s1 = st.selectbox("选择测试集比例", num_s)
RS1 = st.text_input("分类随机种子默认0", 0)
cs = st.selectbox("选择分类", ("是", "否"))
if cs =="是":
from sklearn.model_selection import train_test_split
train_feature, test_feature, y_train, y_test = train_test_split(feature, model.labels_,
test_size=num_s1,
random_state=int(RS1))
st.write("模型分类",train_feature.shape, test_feature.shape)
C = st.selectbox("选择分类方法", ("决策树", "KNN","朴素贝叶斯"))
if C =="决策树":
from sklearn.tree import DecisionTreeClassifier
model = DecisionTreeClassifier() # 构建模型
model.fit(train_feature, y_train) # 模型训练
elif C =="KNN":
C_KNN = st.select_slider("KNN数据数量", options=[100,500,1000,2000,5000,10000])
n_neighbors = st.select_slider("n_neighbors数量默认5",options=[2,3,5, 10, 15,20])
from sklearn.neighbors import KNeighborsClassifier
model = KNeighborsClassifier(n_neighbors=n_neighbors)
model.fit(train_feature[:C_KNN], y_train[:C_KNN]) # 模型训练
elif C =="朴素贝叶斯":
from sklearn.naive_bayes import GaussianNB
model = GaussianNB() # 构建模型
model.fit(train_feature, y_train) # 模型训练
else:
pass
y_pre = model.predict(test_feature)
pre = st.select_slider("控制显示数量", options=[20, 50, 100])
st.write(y_pre[0:pre])
from sklearn.metrics import precision_score, recall_score
from sklearn.metrics import f1_score, confusion_matrix
st.write('模型的精度为：', precision_score(y_test, y_pre, average='macro'))
st.write('模型的召回率为：', recall_score(y_test, y_pre, average='macro'))
st.write('模型的混淆矩阵为：\n', confusion_matrix(y_test, y_pre))
from sklearn.metrics import classification_report
st.write('分类模型的性能评估报告：\n', classification_report(y_test, y_pre))
else:
pass
def ALG(ALG):
if ALG == "机器学习模型训练":
st.write(ALG)
read = st.sidebar.selectbox("读取文件类型", ("csv", "excle"))
title = st.sidebar.text_input('输入地址', 'E:/桌面/项目实训/电商优惠券特征数据')
flie = pd.read_csv(title+"."+read, encoding="GBK")
n = st.select_slider("滑动显示数量", options=[5, 10, 20, 50, 100])
show = st.selectbox("是否显示表", ("否", "是"))
if show == "否":
st.write("不显示")
elif show == "是":
st.dataframe(flie[0:n])
else:
pass
Fl = st.multiselect("选择特征列",flie.columns)
if Fl !=[]:
choose = flie[Fl]
st.write(choose.describe())
from sklearn.preprocessing import StandardScaler, MinMaxScaler
md = st.selectbox("数据标准化", ("标准差标准化StandardScaler"
, "最大最小标准化MinMaxScaler"))
if md == "标准差标准化StandardScaler":
scale = StandardScaler()
elif md == "最大最小标准化MinMaxScaler":
scale = MinMaxScaler()
else:
pass
cf = st.selectbox("标准化", ("是", "否"))
if cf =="是":
feature = scale.fit_transform(choose)
st.write("均值：",feature.mean())
st.write("方差：",feature.std())
from sklearn.cluster import KMeans
num_c = list(range(2, 12))
num_c1 = st.selectbox("选择聚类数量", num_c)
RS = st.text_input("聚类随机种子默认0",0)
cf = st.selectbox("选择聚类",("否","是"))
if cf =="是":
model = KMeans(n_clusters=int(num_c1), random_state=int(RS))
model.fit(feature)
st.write("聚类中心",model.cluster_centers_)
labels = st.select_slider("显示数量", options=[20, 50, 100])
st.write("聚类标签",model.labels_[0:labels])
st.write("聚类标签与数量",pd.Series(model.labels_).value_counts())
def radarplot(model_center=None, label=None):
import matplotlib.pyplot as plt
plt.rcParams['font.sans-serif'] = ['SimHei'] # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False # 用来正常显示负号
n = len(label) # 特征个数
# 对labels进行封闭，否则会有因为matplotlib版本引起的错误
label = np.concatenate((label, [label[0]]))
# 间隔采样，设置雷达图的角度，用于平分切开一个圆面,endpoint设置为False表示随机采样不包括stop的值
angles = np.linspace(0, 2 * np.pi, n, endpoint=False)
# 拼接多个数组，使雷达图一圈封闭起来
angles = np.concatenate((angles, [angles[0]]))
# 创建一个空白画布
fig = plt.figure(figsize=(8, 8))
fig, ax = plt.subplots()#这个语句是用来传递图形的
# 创建子图，设置极坐标格式，绘制圆形
ax = fig.add_subplot(1, 1, 1, polar=True)
# 添加每个特征的标签
ax.set_thetagrids(angles * 180 / np.pi, label)
# 设置y轴范围
ax.set_ylim(model_center.min(), model_center.max())
# 添加网格线
ax.grid(True)
# 设置备选的折线颜色和样式,防止线条重复
sam = ['r', 'pink', 'g', 'b', 'm', 'y', 'k', '0', 'c']
mak = ['4', '8', 'x', '*', 'd', '_', '.', '+', '|']
labels = []
# 循环添加每个类别的线圈
for i in range(len(model_center)):
values = np.concatenate((model_center[i], [model_center[i][0]]))
ax.plot(angles, values, c=sam[i], marker=mak[i])
plt.yticks(fontsize=15)
plt.xticks(fontsize=15)
labels.append(str(i)+"类")
plt.legend(labels, fontsize=15)
st.pyplot(fig)
show_radarplot = st.selectbox("是否显示雷达图", ("是", "否"))
if show_radarplot =="是":
radarplot(model.cluster_centers_, choose.columns)
else:
pass
num_s = list([0.1,0.15,0.2, 0.25,0.3,0.35, 0.4,0.45, 0.5, 0.55,0.6,0.65, 0.7,0.75, 0.8,0.85, 0.9])
num_s1 = st.selectbox("选择测试集比例", num_s)
RS1 = st.text_input("分类随机种子默认0", 0)
cs = st.selectbox("选择分类", ("是", "否"))
if cs =="是":
from sklearn.model_selection import train_test_split
train_feature, test_feature, y_train, y_test = train_test_split(feature, model.labels_,
test_size=num_s1,
random_state=int(RS1))
st.write("模型分类",train_feature.shape, test_feature.shape)
C = st.selectbox("选择分类方法", ("决策树", "KNN","朴素贝叶斯"))
if C =="决策树":
from sklearn.tree import DecisionTreeClassifier
model = DecisionTreeClassifier() # 构建模型
model.fit(train_feature, y_train) # 模型训练
elif C =="KNN":
C_KNN = st.select_slider("KNN数据数量", options=[100,500,1000,2000,5000,10000])
n_neighbors = st.select_slider("n_neighbors数量默认5",options=[2,3,5, 10, 15,20])
from sklearn.neighbors import KNeighborsClassifier
model = KNeighborsClassifier(n_neighbors=n_neighbors)
model.fit(train_feature[:C_KNN], y_train[:C_KNN]) # 模型训练
elif C =="朴素贝叶斯":
from sklearn.naive_bayes import GaussianNB
model = GaussianNB() # 构建模型
model.fit(train_feature, y_train) # 模型训练
else:
pass
y_pre = model.predict(test_feature)
pre = st.select_slider("控制显示数量", options=[20, 50, 100])
st.write(y_pre[0:pre])
from sklearn.metrics import precision_score, recall_score
from sklearn.metrics import f1_score, confusion_matrix
st.write('模型的精度为：', precision_score(y_test, y_pre, average='macro'))
st.write('模型的召回率为：', recall_score(y_test, y_pre, average='macro'))
st.write('模型的混淆矩阵为：\n', confusion_matrix(y_test, y_pre))
from sklearn.metrics import classification_report
st.write('分类模型的性能评估报告：\n', classification_report(y_test, y_pre))
else:
pass以往的文章中有streamlit的通用模板，包括登录界面的实现还有框架解析
在做的过程中，发现streamlit可以搭载在wifi（局域网上）可以通过局域网打开streamlit达到共享的作用。
扫一扫
2253
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
