https://haosquare.com/tags/data-analytics/Recent content in Data-Analytics on 好豪筆記Hugo -- gohugo.iozh-twhaosquare.tw@gmail.com (好豪)haosquare.tw@gmail.com (好豪)2026 好豪筆記 ʕ •ᴥ•ʔThu, 25 Dec 2025 00:00:00 +0000https://haosquare.com/zero-tech-communication-strategy-2025/Thu, 25 Dec 2025 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/zero-tech-communication-strategy-2025/同一份分析報告、對不同聽眾的三場簡報分享，我學到：影響力不只是用技術深度換來的，更需要用對的語言、向對的人、說對的事。https://haosquare.com/mix-shift-composition-effect/Mon, 17 Nov 2025 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/mix-shift-composition-effect/留存率下降、流失率上升？不見得是產品出問題！用 Mix-Shift 分辨是「產品真的有問題」還是「用戶組成改變」https://haosquare.com/sequential-ab-testing/Sat, 01 Nov 2025 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/sequential-ab-testing/Sequential A/B Testing 不止允許偷看數據、需要的樣本數比標準假設檢定還少！白話介紹這個 Netflix、Spotify 都在用的神奇方法https://haosquare.com/2025-instagram-youtube-announcement/Thu, 06 Feb 2025 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/2025-instagram-youtube-announcement/歡迎追蹤我的 Instagram 帳號 @hao2huanghttps://haosquare.com/r-vs-python-data-science/Sun, 12 Jan 2025 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/r-vs-python-data-science/比起 R 語言，我建議以台灣資料科學職場為目標的新手先專注 Pythonhttps://haosquare.com/normal-distribution-central-limit-theorem-t-test/Sun, 10 Nov 2024 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/normal-distribution-central-limit-theorem-t-test/常態分佈是一個非常基礎但又容易被誤解的概念，只要把各個知識點連結起來，其實沒那麼複雜https://haosquare.com/decipher-openai-data-scientist-career/Sun, 28 Jul 2024 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/decipher-openai-data-scientist-career/ChatGPT 崛起對資料科學家職涯造成什麼衝擊？數據分析技能會被 AI 取代嗎？https://haosquare.com/north-star-metric-linkedin-data-intuition-framework/Mon, 22 Jul 2024 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/north-star-metric-linkedin-data-intuition-framework/LinkedIn 定義北極星指標的方法、以及如何驅動產品進步https://haosquare.com/udemy-naive-bayes-intro/Sun, 14 Jul 2024 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/udemy-naive-bayes-intro/儘管深度學習模型光彩奪目，Naive Bayes 依然因為它簡單、高效、好解釋而有其地位https://haosquare.com/ab-test-power-analysis-intuition-busters/Sun, 26 May 2024 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/ab-test-power-analysis-intuition-busters/統計檢定力 Power Analysishttps://haosquare.com/data-scientist-2024-job-interview/Sun, 19 May 2024 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/data-scientist-2024-job-interview/2024 年面試資深資料科學家與數據分析師的經驗分享，以亞洲地區的大科技公司為主https://haosquare.com/ab-test-significance-level-false-positive-rate/Sun, 12 May 2024 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/ab-test-significance-level-false-positive-rate/介紹偽陽性率，它與顯著水準、統計檢定力、以及決策品質的關聯https://haosquare.com/survivorship-bias-and-right-censored-data/Mon, 22 Apr 2024 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/survivorship-bias-and-right-censored-data/分享極其常見的倖存者偏差、以及與其相似的「死者偏誤」與右設限資料https://haosquare.com/chi-squared-test-application/Sun, 14 Apr 2024 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/chi-squared-test-application/資料科學家不得不知道的三大卡方檢定使用時機：多組別的 A/B 測試、評估二分類模型、以及機器學習模型特徵選擇https://haosquare.com/data-science-tech-company-blog/Tue, 19 Mar 2024 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/data-science-tech-company-blog/閱讀技術部落格是成為更強資料科學家必經之路https://haosquare.com/type-i-error-and-type-ii-error/Sat, 09 Dec 2023 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/type-i-error-and-type-ii-error/如果你還無法搞懂 P 值、顯著水準、或檢定力，型一與型二錯誤正是它們共通的重要觀念，這篇文章將用圖示讓你輕鬆理解！https://haosquare.com/sql-window-function-intro/Wed, 11 Oct 2023 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/sql-window-function-intro/這篇教學將讓你在 2 分鐘內快速認識 SQL 窗口函數好用之處，學會這項資料科學家必備技能！https://haosquare.com/chi-squared-test/Wed, 04 Oct 2023 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/chi-squared-test/白話介紹三種最常見的卡方檢定：配適度檢定、獨立性檢定、同質性檢定，教學如何用 Python 的 Scipy 套件來實作https://haosquare.com/roc-curve-best-cutoff/Sat, 23 Sep 2023 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/roc-curve-best-cutoff/3 種在 ROC 曲線上選擇最佳閾值的方法，包括 Youden’s J statistic、最近距離法、還有一種用肉眼就能看出最佳閾值的平行線判別法https://haosquare.com/roc-curve/Fri, 15 Sep 2023 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/roc-curve/讀完這篇將再也不會忘記 ROC 曲線、TPR、FPR 的定義！https://haosquare.com/calculate-ab-testing-sample-size/Tue, 27 Jun 2023 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/calculate-ab-testing-sample-size/A/B Test 樣本數計算的公式與假設https://haosquare.com/find-data-science-internship/Fri, 16 Jun 2023 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/find-data-science-internship/分享我在大學畢業以前找到三份實習的經驗，一一解析從尋找職缺、整理履歷、到挑戰面試的各階段該怎麼準備才更有效率https://haosquare.com/one-tailed-vs-two-tailed-test/Sun, 05 Mar 2023 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/one-tailed-vs-two-tailed-test/簡短說明單尾與雙尾檢定的統計學特性差異，也介紹單尾檢定常見的錯誤使用方法https://haosquare.com/why-bayeisan-statistics/Sat, 26 Nov 2022 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/why-bayeisan-statistics/貝氏統計幫助你在模型加入個人觀點、讓統計模型隨著新的資料取得不斷進化、並且量化你對決策究竟有多麽「不確定」https://haosquare.com/calling-bullshit/Sat, 12 Nov 2022 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/calling-bullshit/到底這世上為什麼數據鬼扯如此氾濫？！https://haosquare.com/why-google-colab-python/Sat, 06 Aug 2022 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/why-google-colab-python/Google Colab 是讓你用網頁瀏覽器就能寫 Python 程式並執行的超讚平台https://haosquare.com/ab-testing-novelty-primacy-effect/Sun, 03 Jul 2022 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/ab-testing-novelty-primacy-effect/A/B 測試不得不知道的 Novelty 與 Primacy Effect 問題，了解顧客的「好奇」與「守舊」傾向影響 A/B Testing 判讀https://haosquare.com/python-pandas-lambda-tricks/Fri, 03 Jun 2022 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/python-pandas-lambda-tricks/五個案例分享 Pandas 內 Lambda 函式的超實用使用技巧https://haosquare.com/standard-error/Sun, 15 May 2022 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/standard-error/標準誤與標準差有什麼差別？https://haosquare.com/causal-inference-intro/Sun, 13 Mar 2022 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/causal-inference-intro/簡介因果推論（Causal Inference）這個研究領域，用範例說明分析因果關係的三大困難點：混淆變項、自我選擇偏誤、以及反事實https://haosquare.com/python-pandas-method-chaining/Sat, 05 Mar 2022 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/python-pandas-method-chaining/Pandas 的 Method Chaining 實現「一行只有一個資料處理動作」的程式碼寫作風格，讓你的 Pandas 程式碼更簡潔、更好讀https://haosquare.com/python-pandas-wide-and-long-data/Tue, 01 Feb 2022 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/python-pandas-wide-and-long-data/Python 的 Pandas 套件裡面好用的長寬資料轉換函式：stack()、unstack()、melt()、pivot()https://haosquare.com/confidence-interval/Thu, 20 Jan 2022 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/confidence-interval/你知道信賴區間的 95% 機率是指什麼嗎？標準差跟標準誤哪裡不一樣？https://haosquare.com/data-scientist-2021-job-interview/Wed, 29 Dec 2021 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/data-scientist-2021-job-interview/橫跨新加坡、泰國、還有台灣的外商資料科學家面試經驗https://haosquare.com/multiple-testing-intro/Mon, 06 Dec 2021 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/multiple-testing-intro/分享什麼是多重檢定問題，他到底跟樂透、A/B Test、或者探索性資料分析有什麼關聯，並且介紹用來解決多重檢定問題的 Bonferroni 校正https://haosquare.com/hypothesis-testing-for-proportion/Fri, 06 Aug 2021 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/hypothesis-testing-for-proportion/介紹 Z 檢定與卡方檢定之間的關聯，以及什麼情況下這兩者都可以使用https://haosquare.com/r-learning-resources/Wed, 21 Jul 2021 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/r-learning-resources/R 語言適合自學的教材，包括書籍、線上課程、以及練習題，希望讓更多人知道 R 語言厲害之處！https://haosquare.com/aa-testing-before-ab-testing/Sat, 17 Jul 2021 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/aa-testing-before-ab-testing/A/A Test 跟 A/B Test 不一樣，A/A Test 是 A/B Test 上線執行以前的重要前置準備https://haosquare.com/covid-19-data-viz/Tue, 20 Oct 2020 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/covid-19-data-viz/我認為這張疫情圖表是最佳的資料視覺化範例之一，它包含《Google 必修的圖表簡報術》這本書提及最重要的幾項技巧https://haosquare.com/user-cycle/Fri, 04 Sep 2020 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/user-cycle/用數據分析與開放資料告訴你如何編排與客戶互動的時機https://haosquare.com/ab-testing-peeking/Sat, 22 Aug 2020 00:00:00 +0000haosquare.tw@gmail.com (好豪)https://haosquare.com/ab-testing-peeking/在實驗樣本蒐集完整之前「偷看結果」，最嚴重的下場，會讓你手中看似有統計顯著性的決策，20% 以上是錯誤的！