diff --git a/Chapter_3_GettingStarted/Baseline_RealWorldData.ipynb b/Chapter_3_GettingStarted/Baseline_RealWorldData.ipynb
index 4949bf0..0601e1e 100644
--- a/Chapter_3_GettingStarted/Baseline_RealWorldData.ipynb
+++ b/Chapter_3_GettingStarted/Baseline_RealWorldData.ipynb
@@ -10,7 +10,7 @@
     "\n",
     "The previous section illustrated how simple data preprocessing and supervised learning techniques can be used to design a baseline fraud detection system. The presented results relied on reproducible, but simulated data. Let us now apply the exact same methodology with real-world transaction data. Due to confidentiality reasons, the data used in this section cannot be shared. While the results presented in this section cannot be reproduced, they however provide insights into the performances that would be obtained in a real-world setting. \n",
     "\n",
-    "The dataset used was provided By Worldline, and is similar in nature to the datasets used in the publications referenced on our [ResearchGate page - Joint collaboration: MLG ULB and Worldline](https://www.researchgate.net/project/Fraud-detection-with-machine-learning).\n",
+    "The dataset used was provided by Worldline, and is similar in nature to the datasets used in the publications referenced on our [ResearchGate page - Joint collaboration: MLG ULB and Worldline](https://www.researchgate.net/project/Fraud-detection-with-machine-learning).\n",
     "\n",
     "## Training and test sets\n",
     "\n",