{"id":2497,"date":"2026-02-02T08:33:14","date_gmt":"2026-02-02T08:33:14","guid":{"rendered":"https:\/\/demo.materiamedica.net\/demo6\/?p=2497"},"modified":"2026-02-02T08:33:14","modified_gmt":"2026-02-02T08:33:14","slug":"chapter-1-random-numbers-in-numpy","status":"publish","type":"post","link":"https:\/\/demo.materiamedica.net\/demo6\/chapter-1-random-numbers-in-numpy\/","title":{"rendered":"Chapter 1: Random Numbers in NumPy"},"content":{"rendered":"

Random Numbers in NumPy<\/strong> \u2014 written as if I\u2019m your patient teacher sitting next to you, showing examples on the screen, explaining every important detail, giving realistic use-cases, warning about common mistakes, and helping you build good habits from the beginning.<\/p>\n

Let\u2019s imagine we\u2019re working together in a Jupyter notebook. Ready? \ud83d\ude0a<\/p>\n

\n
\n
\n
Python<\/div>\n
\n
import numpy as np<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
1. Why NumPy random instead of Python\u2019s random?<\/h3>\n
Many beginners start with Python\u2019s built-in random module.<\/p>\n
But in serious numerical\/scientific\/ML work you should almost always<\/strong> use numpy.random.<\/p>\n
Main reasons:<\/p>\n
    \n
  • Much faster<\/strong> when generating thousands\/millions of numbers<\/li>\n
  • Returns NumPy arrays<\/strong> directly (ready for math, slicing, broadcasting)<\/li>\n
  • Much larger variety<\/strong> of distributions (normal, binomial, poisson, beta, gamma, multivariate\u2026)<\/li>\n
  • Proper multi-dimensional support<\/strong> out of the box<\/li>\n
  • Consistent seeding<\/strong> behavior across different functions<\/li>\n
  • Better integration with the rest of the NumPy ecosystem<\/li>\n<\/ul>\n
    Golden rule #1 (write it somewhere):<\/strong><\/p>\n
    \n
    If you’re doing anything numerical\/scientific\/ML\/data analysis \u2192 use numpy.random<\/strong> \u2014 not random, not secrets, not anything else.<\/p>\n<\/blockquote>\n
    2. The MOST important habit: Always control the seed<\/h3>\n
    \n
    \n
    \n
    Python<\/div>\n
    \n
    np.random.seed(42)          # \u2190 this single line changes everything<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
    Why is this so important?<\/strong><\/p>\n
      \n
    • Makes experiments reproducible<\/strong><\/li>\n
    • Makes debugging much easier<\/strong><\/li>\n
    • Allows others to get exactly the same results<\/strong><\/li>\n
    • Extremely helpful when teaching, writing tutorials, comparing models<\/li>\n<\/ul>\n
      Quick demonstration:<\/strong><\/p>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      # Without seed \u2192 different every time you run\r\nprint(np.random.rand(5))\r\n\r\n# With seed \u2192 always exactly the same numbers\r\nnp.random.seed(42)\r\nprint(np.random.rand(5))\r\n# \u2192 [0.37454012 0.95071431 0.73199394 0.59865848 0.15601864]<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      Teacher tip<\/strong>: During learning\/experimenting \u2192 always<\/strong> put a fixed seed at the top of your notebook\/script. When you go to production or want true randomness \u2192 remove\/comment out the seed line.<\/p>\n
      3. The most frequently used random functions (you’ll use these 90% of the time)<\/h3>\n
      A. np.random.rand() \u2014 Uniform random numbers [0, 1)<\/h4>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      # Single number\r\nprint(np.random.rand())\r\n\r\n# 1D vector\r\nprint(np.random.rand(10))\r\n\r\n# 2D matrix (very common!)\r\nprint(np.random.rand(4, 6))\r\n\r\n# 4D \u2014 batch of images (common in deep learning)\r\nprint(np.random.rand(32, 64, 64, 3).shape)  # (32, 64, 64, 3)<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      B. np.random.randn() \u2014 Standard normal (Gaussian) distribution<\/h4>\n
      Mean = 0, standard deviation = 1<\/p>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      print(np.random.randn(12))\r\n# typical output: mixture of positive & negative, most values between -3 and +3\r\n\r\n# Very common shape in neural networks\r\nweights_layer1 = np.random.randn(784, 128)   # 784 input \u2192 128 neurons<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      Quick comparison table students should remember:<\/strong><\/p>\n
      \n
      \n\n\n\n\n\n\n\n
      Function<\/th>\nRange \/ Distribution<\/th>\nTypical shape example<\/th>\nMost common use case<\/th>\n<\/tr>\n<\/thead>\n
      rand()<\/td>\n[0, 1) uniform<\/td>\nrand(1000, 30)<\/td>\nFeatures, dropout masks, probabilities<\/td>\n<\/tr>\n
      randn()<\/td>\n~ Normal(0, 1)<\/td>\nrandn(512, 512, 3)<\/td>\nWeights init, noise, synthetic data<\/td>\n<\/tr>\n
      randint()<\/td>\nintegers [low, high)<\/td>\nrandint(1, 7, size=100)<\/td>\nDice, labels, indices, pixel values<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
      <\/div>\n<\/div>\n<\/div>\n
      C. np.random.randint(low, high, size=…) \u2014 Random integers<\/h4>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      # Single dice roll (1\u20136)\r\nprint(np.random.randint(1, 7))\r\n\r\n# 1000 coin flips (0 or 1)\r\ncoins = np.random.randint(0, 2, size=1000)\r\n\r\n# Realistic: exam scores (0\u2013100)\r\nscores = np.random.randint(35, 101, size=(200, 5))  # 200 students \u00d7 5 subjects<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      D. np.random.uniform(low, high, size=…) \u2014 Uniform in custom range<\/h4>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      # Temperatures between 18\u201332 \u00b0C\r\ntemps = np.random.uniform(18, 32, size=365)\r\n\r\n# Prices between 9.99 and 99.99\r\nprices = np.random.uniform(9.99, 99.99, size=500)<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      E. np.random.normal(loc, scale, size=…) \u2014 Custom normal distribution<\/h4>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      # Realistic IQ scores\r\niq = np.random.normal(loc=100, scale=15, size=10000)\r\n\r\n# Measurement with sensor noise (\u00b12 units)\r\ntrue_distance = 150.0\r\nmeasured = true_distance + np.random.normal(0, 2, size=200)<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      4. Very common realistic use cases (you will write these many times)<\/h3>\n
      Use case 1 \u2013 Synthetic training data<\/strong><\/p>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      np.random.seed(42)\r\n\r\nn = 5000\r\nX = np.random.randn(n, 8) * 3 + 50               # centered around 50\r\nnoise = np.random.normal(0, 4, n)\r\n\r\ny = (2.5 * X[:,0] - 1.8 * X[:,2] + 0.9 * X[:,5]) + noise<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      Use case 2 \u2013 Adding realistic noise to images<\/strong><\/p>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      img = np.full((300, 400, 3), 180, dtype=np.uint8)  # light gray\r\nnoise = np.random.normal(0, 30, img.shape)\r\nnoisy = np.clip(img + noise, 0, 255).astype(np.uint8)<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      Use case 3 \u2013 Random train\/val\/test split (manual)<\/strong><\/p>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      data = np.random.randn(12000, 25)\r\n\r\nnp.random.shuffle(data)          # \u2190 very important step!\r\n\r\ntrain = data[:9000]\r\nval   = data[9000:10500]\r\ntest  = data[10500:]<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      Use case 4 \u2013 Random sampling \/ bootstrapping<\/strong><\/p>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      population = np.random.normal(100, 20, 100000)\r\n\r\n# Bootstrap sample of 1000\r\nsample = np.random.choice(population, size=1000, replace=True)<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      Summary \u2013 Your Quick Reference Table<\/h3>\n
      \n
      \n\n\n\n\n\n\n\n\n\n\n\n
      Function<\/th>\nWhat you get<\/th>\nTypical shape<\/th>\nTypical seed usage<\/th>\n<\/tr>\n<\/thead>\n
      rand()<\/td>\nUniform [0,1)<\/td>\nrand(1000,20)<\/td>\nYes<\/td>\n<\/tr>\n
      randn()<\/td>\nStandard normal (\u03bc=0, \u03c3=1)<\/td>\nrandn(512,512,3)<\/td>\nYes<\/td>\n<\/tr>\n
      randint(low,high)<\/td>\nIntegers [low, high)<\/td>\nrandint(1,101,10000)<\/td>\nYes<\/td>\n<\/tr>\n
      uniform(a,b)<\/td>\nUniform [a,b)<\/td>\nuniform(0,100,500)<\/td>\nYes<\/td>\n<\/tr>\n
      normal(\u03bc,\u03c3)<\/td>\nNormal distribution \u03bc,\u03c3<\/td>\nnormal(170,10,10000)<\/td>\nYes<\/td>\n<\/tr>\n
      choice()<\/td>\nSample from given array<\/td>\nchoice(names, 200)<\/td>\nYes<\/td>\n<\/tr>\n
      shuffle()<\/td>\nShuffle array in place<\/strong><\/td>\nshuffle(dataset)<\/td>\nYes (before)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
      <\/div>\n<\/div>\n<\/div>\n
      Final teacher advice (very important)<\/h3>\n
      Always<\/strong> start your random-related work with:<\/p>\n
      \n
      \n
      \n
      Python<\/div>\n
      \n
      import numpy as np\r\nnp.random.seed(42)           # or 123, 0, 777 \u2014 any fixed number<\/code><\/pre>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n
      This tiny habit will save you many hours of confusion later.<\/p>\n
      Where do you want to go next?<\/p>\n
        \n
      • More exotic distributions (poisson, binomial, beta, gamma\u2026)<\/li>\n
      • Randomness in machine learning (dropout, weight init, data augmentation)<\/li>\n
      • Common bugs & misunderstandings with random numbers<\/li>\n
      • Mini-project: generate synthetic dataset + add noise + visualize<\/li>\n
      • Reproducibility tricks across multiple files\/notebooks<\/li>\n<\/ul>\n
        Just tell me what feels most interesting or useful for you right now! \ud83d\ude80<\/p>\n","protected":false},"excerpt":{"rendered":"
        Random Numbers in NumPy \u2014 written as if I\u2019m your patient teacher sitting next to you, showing examples on the screen, explaining every important detail, giving realistic use-cases, warning about common mistakes, and helping...<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[75],"tags":[],"class_list":["post-2497","post","type-post","status-publish","format-standard","hentry","category-numpy"],"_links":{"self":[{"href":"https:\/\/demo.materiamedica.net\/demo6\/wp-json\/wp\/v2\/posts\/2497","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/demo.materiamedica.net\/demo6\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/demo.materiamedica.net\/demo6\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/demo.materiamedica.net\/demo6\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/demo.materiamedica.net\/demo6\/wp-json\/wp\/v2\/comments?post=2497"}],"version-history":[{"count":1,"href":"https:\/\/demo.materiamedica.net\/demo6\/wp-json\/wp\/v2\/posts\/2497\/revisions"}],"predecessor-version":[{"id":2498,"href":"https:\/\/demo.materiamedica.net\/demo6\/wp-json\/wp\/v2\/posts\/2497\/revisions\/2498"}],"wp:attachment":[{"href":"https:\/\/demo.materiamedica.net\/demo6\/wp-json\/wp\/v2\/media?parent=2497"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/demo.materiamedica.net\/demo6\/wp-json\/wp\/v2\/categories?post=2497"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/demo.materiamedica.net\/demo6\/wp-json\/wp\/v2\/tags?post=2497"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}