- 数据预处理Data preprocessing
缺失值处理Missing value processing(插值填补Interpolation filling,多重填补multiple filling、miceforest算法填补)
异常值处理Outlier processing(直接剔除Direct culling,3σ盖帽法Three sigma capping method,四分位点法Quartile point method,绝对中位差算法Absolute median difference algorithm)
- 特征工程Feature Engineering
特征选择feature selection(方差选择法Variance selection method(剔除方差小于某一阈值的列Remove columns whose variance is less than a certain threshold),树模型选择法Tree model selection method)
Autoencoder自编码器
降维方法descending dimension method
(PCA降维PCA dimensional reduction)
- 机器学习模型Machine learning model
分类模型Classification model:
(logistic分类,决策树(Decision Tree),随机森林(Random forest),SVM支持向量机(Support Vector Machine),xgboost分类,神经网络分类Neural network classification)
回归模型Regression models:
(高斯过程回归(Gaussian Process Regression, GPR),xgboost回归,线性回归(linear regression),神经网络回归)-*
聚类Clustering model:
谱聚类(spectral clustering),k-means++ clustering,层次聚类(hierarchical clustering),Density-Based Spatial Clustering of Applications with Noise(DBSCAN密度聚类)
时间序列预测Time series forecasting:
LSTM( Long Short Term Memory Neural networks长短期记忆神经网络)
RNN(Recurrent Neural Network循环神经网络)
ARIMA(Autoregressive Integrated Moving Average Model)
SARIMA(Seasonal Autoregressive Integrated Moving Average Model)
关联规则挖掘(association rule mining):
Apriori、FP-Growth算法algorithm
四、微分方程模型(differential equation model)
Logistic模型
传染病模型Mathematical models of epidemic diseases(如SI、SIR、SIRS、SEIR 模型)
人口增长模型(Population growth model)
- 优化模型optimization model
作业车间调度问题(Job Shop Scheduling, JSP)
TSP问题(Travelling Salesman Problem)
装箱问题(bin-packing problem)
选址问题Location problem
最短路问题Shortest path problem
多目标规划MOP(multi-objective programming)
- 综合评价模型Comprehensive evaluation model:
TOPSIS (Technique for Order Preference by Similarity to an Ideal Solution )
层次分析法Analytic Hierarchy Process
秩和比法(Rank-sum ratio,简称RSR法)
灰色关联度分析(Grey Relation Analysis,GRA)
熵权法(the entropy weight method 简称EWM)
- 常用算法common algorithms:
解决优化模型的算法Algorithm to solve the optimization model:
模拟退火算法(Simulated Annealing,SA)
遗传算法(Genetic Algorithm,GA)
人工蜂群算法(Artificial Bee Colony Algorithm, 简称ABC算法)
人工鱼群算法artificial fish swarms algorithm (AFSA)
梯度下降(Gradient Descent)
反向传播算法Backpropagation algorithm
龙格-库塔法(Runge-Kutta methods)
谱聚类(spectral clustering),k-means++ clustering,层次聚类(hierarchical clustering),Density-Based Spatial Clustering of Applications with Noise(DBSCAN密度聚类)
常用词汇:序列(sequence),递归(recursion),递归神经网络(recursive neural network),双向循环神经网络(Bidirectional RNN, Bi-RNN),图灵完备(Turing completeness),自然语言处理(Natural Language Processing, NLP),卷积神经网络(Convolutional Neural Network,CNN),隐含层(hidden layer),时间步(time-step),归一化指数函数(softmax function),监督学习(supervised learning),非监督学习(unsupervised learning),梯度消失(gradient vanishing)和梯度爆炸(gradient explosion),权重初始化(weight initialization),激活函数(Activation Function),Tanh 激活函数又叫作双曲正切激活函数(hyperbolic tangent activation function),ReLU函数又称为修正线性单元(Rectified Linear Unit),预训练模型(Pre-train Model),语言模型(LM: Language Model),自回归(AR: Auto-Regressive),自编码(AE: Auto-Encoding)。卷积神经网络中提及的神经认知机(neocognitron),卷积神经网络中卷积层(convolution layer)和池化层(pooling layer),隐马尔可夫模型(Hidden Markov Model, HMM),随机梯度下降(Stochastic Gradient Descent, SGD),反向传播算法(Back-Propagation, BP),偏差量(bias vector),神经元(neuron),感受野(receptive field),非线性偏微分方程(Nonlinear partial differential equation),扩散方程 (Diffusion Equation),方程(equation)