Skip to content. | Skip to navigation

Personal tools


You are here: Home / Tutorial / Math


Statistics, Algorithm, Dataset 統計重點在於分析因果關係, 機器學習重點在於進行預測 numpy迴圈加速

分組數據: 眾數=出現頻數最多的數值; 中位數=總頻數除2 例17.5代表第17和第18的數值相加除2; 平均數; 標準差

統計學知識 圓錐切面

高中生的神經網路入門 Big O Notation and Time Complexity BigO Notation decision-trees-an-intuitive-introduction poisson-distribution-and-poisson-process-explained

SageMath 使用 Python 語法,可直接在瀏覽器操作,偏向計算功能而不像 Geogebra 視覺化豐富,適合當高中生的數學計算輔助工具。

Data Science from Scratch: Finding Key Connectors

PCA: 當資料維度數 (變數) 很多的時候, 想辦法讓維度降低, 但資料特性不會差太多

Visualizing Principal Component Analysis with Matrix Transforms: Principal Component Analysis (PCA) is a method of decomposing data into uncorrelated components by identifying eigenvalues and eigenvectors

mathematics-for-data-science Math Adventures with Python improve-your-data-structures-algorithms-and-problem-solving-skills 因數 質數 Prime Number Generator行列式是為了解方程式,陣列可以方便地解題。 用數學講故事: Neal Stephenson Cryptonomicon

import string, random

def permute(pool):
    Randomly permute whatever list is passed in as pool
    oldlist = []+list(pool)
    newlist = []
    for i in range(len(oldlist)): # = range(26)
        randitem = random.choice(oldlist)
    return newlist

Statistics for Data Scientists hidden-markov-model-part-1 轉移矩陣 狀態矩陣


Probability Fundamentals of Machine Learning Knapsack Problem 背包問題 固定金額購買股票

NumPy 執行效率最高 1秒的話 ListComp 是 80秒 Map 是 100秒 for 迴圈是 180秒





MNIST from local files


Logistic: 常用於 Binary Classification 預測訪客是否會點擊某個廣告連結 Binary and Multiclass Logistic Regression with GD and Newton’s Method

因為 Linear Regression 的結果並不在 0 與 1 之間, 難以表示成 Y=0 或 Y=1 的機率分佈; Sigmoid Function 將數值對應到 0 與 1 之間

Bayesian Estimation

推薦系統 Star Ratings: Bayesian Average

Practical Data Science Cookbook: Recommending Movies at Scale (Python), Harvesting and Geolocating Twitter Data (Python), Optimizing Numerical Code with NumPy and SciPy

Bayes theorem the holy grail of Data Science interpreting-complex-models-with-shap-values how-to-stop-worrying-and-start-writing-tests-based-on-properties

Machine Learning

Feature Selection with sklearn and Pandas 3 primary types of Gradient Descent Support Vector Machine SVM 在地上修一條公路把不同類型的點分開 並讓這條路越寬越好 Introduction to k-Nearest Neighbors nearest neighbors with keras and coreml


explaining supervised learning to a kid introduction-to machine learning algorithms linear regression custom tensorflow loss functions for advanced machine learning the hottest topics in machine learning top sources for machine learning datasets build your own neural network-1-the-story-of-perceptron-sigmoid visual-machine-learning-for-artists is-learning-rate-useful-in-artificial-neural-networks machine-learning-vs-deep-learning-image-classification neural-networks-and-backpropagation-machine-learning-made-easy identifying-the-genre-of-a-song with neural networks bag-of-tricks for image classification with convolutional neural networks in keras identifying the genre of a song with neural networks what-are-eigenvalues-and-eigenvectors a must know concept for machine learning

Building CNN in Keras Extreme Rare Event Classification using AutoEncoders in Keras

Tensor 張量: 第0階 rank-0 純量, 第1階 向量, 第2階 矩陣