基于matlab中机器学习工具箱中随机森林模型简单易上手对房价预测实战_房价预测模型matlab

trainingData=data1
inputTable = trainingData;
predictorNames = {'VarName1', 'VarName2', 'VarName3', 'VarName4', 'VarName5', 'VarName6', 'VarName7'};
predictors = inputTable(:, predictorNames);
response = inputTable.m2;
isCategoricalPredictor = [true, false, false, false, true, true, true];
 
% 训练回归模型
% 以下代码指定所有模型选项并训练模型。
template = templateTree(...
    'MinLeafSize', 8);
regressionEnsemble = fitrensemble(...
    predictors, ...
    response, ...
    'Method', 'Bag', ...
    'NumLearningCycles', 30, ...
    'Learners', template);
 
% 使用预测函数创建结果结构体
predictorExtractionFcn = @(t) t(:, predictorNames);
ensemblePredictFcn = @(x) predict(regressionEnsemble, x);
trainedModel.predictFcn = @(x) ensemblePredictFcn(predictorExtractionFcn(x));
 
% 向结果结构体中添加字段
trainedModel.RequiredVariables = {'VarName1', 'VarName2', 'VarName3', 'VarName4', 'VarName5', 'VarName6', 'VarName7'};
trainedModel.RegressionEnsemble = regressionEnsemble;
trainedModel.About = '此结构体是从回归学习器 R2021a 导出的训练模型。';
trainedModel.HowToPredict = sprintf('要对新表 T 进行预测，请使用: \n yfit = c.predictFcn(T) \n将 ''c'' 替换为作为此结构体的变量的名称，例如 ''trainedModel''。\n \n表 T 必须包含由以下内容返回的变量: \n c.RequiredVariables \n变量格式(例如矩阵/向量、数据类型)必须与原始训练数据匹配。\n忽略其他变量。\n \n有关详细信息，请参阅 <a href="matlab:helpview(fullfile(docroot, ''stats'', ''stats.map''), ''appregression_exportmodeltoworkspace'')">How to predict using an exported model</a>。');
 
% 提取预测变量和响应
% 以下代码将数据处理为合适的形状以训练模型。
%
inputTable = trainingData;
predictorNames = {'VarName1', 'VarName2', 'VarName3', 'VarName4', 'VarName5', 'VarName6', 'VarName7'};
predictors = inputTable(:, predictorNames);
response = inputTable.m2;
isCategoricalPredictor = [true, false, false, false, true, true, true];
 
% 执行交叉验证
partitionedModel = crossval(trainedModel.RegressionEnsemble, 'KFold', 5);
 
% 计算验证预测
validationPredictions = kfoldPredict(partitionedModel);
 
% 计算验证 RMSE
validationRMSE = sqrt(kfoldLoss(partitionedModel, 'LossFun', 'mse'));
% 交叉验证后得到的测试集上的衡量指标
isNotMissing = ~isnan(validationPredictions) & ~isnan(response);
format long g  % 数值显示的精度变高
validationSSE = sum(( validationPredictions - response ).^2,'omitnan') ;
validationMSE = sum(( validationPredictions - response ).^2,'omitnan') / numel(response(isNotMissing));
validationRMSE = sqrt(sum(( validationPredictions - response ).^2,'omitnan') / numel(response(isNotMissing) ));
validationMAE = sum(abs( validationPredictions - response),'omitnan') / numel(response(isNotMissing));
validationMAPE = sum(abs((validationPredictions - response)./response),'omitnan') / numel(response(isNotMissing));
validationSMAPE = sum(abs(validationPredictions - response)./((abs(response)+abs(validationPredictions))/2),'omitnan') / numel(response(isNotMissing));
validationR2 = 1 -  sum(( validationPredictions - response ).^2,'omitnan') / sum((response - mean(response,'omitnan')).^2,'omitnan');
validation_result = table(validationSSE,validationMSE,validationRMSE,validationMAE,validationMAPE,validationSMAPE,validationR2)
format short
predict_y = trainedModel.predictFcn(data2)

这样就通过随机森林模型将房价预测好了，但要评价该模型建立的好坏，需要通过R方，即拟合优度\决定系数\可决系数）决定，这里是matlab关于拟合优度的官方链接https://ww2.mathworks.cn/help/stats/coefficient-of-determination-r-squared.html

在随机森林模型中，特征选择也很重要，一些特征可能会影响该模型的效果，故我们选择查看各指标的重要性

importance =oobPermutedPredictorImportance(regressionEnsemble);
 % 可以将重要性归一化到[0 1]区间内  % https://ww2.mathworks.cn/matlabcentral/answers/783611-need-predictor-importance-in-random-forest-expressed-as-a-percentage#answer_671247
importance(importance<0) = 0;
importance = importance./sum(importance)
figure(1)
h_bar = bar(importance);
h_bar.Parent.XTickLabel ={'区域','卧室数','客厅数','房屋面积','楼层高低','是否是地铁房','是否是学区房'};
h_bar.Parent.XTickLabelRotation = 45;  
h_bar.Parent.TickLabelInterpreter =  'none';

效果如图所示

至此，利用matlab的工具箱对房价预测的实战例题已求解完毕