{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.12","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"none","dataSources":[{"sourceId":94147,"databundleVersionId":11390004,"sourceType":"competition"}],"dockerImageVersionId":30918,"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":false}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"

\nEDA on Histology Image Data\n

","metadata":{}},{"cell_type":"code","source":"import h5py\nimport numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nimport seaborn as sns \n\nimport warnings\nwarnings.simplefilter('ignore')","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true,"execution":{"iopub.status.busy":"2025-03-19T11:37:32.074938Z","iopub.execute_input":"2025-03-19T11:37:32.075298Z","iopub.status.idle":"2025-03-19T11:37:32.080057Z","shell.execute_reply.started":"2025-03-19T11:37:32.075271Z","shell.execute_reply":"2025-03-19T11:37:32.078915Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"

\nData Loading and Overview\n

","metadata":{}},{"cell_type":"code","source":"def load_data(file_path):\n with h5py.File(file_path, \"r\") as h5file:\n train_images = {k: np.array(v) for k, v in h5file[\"images/Train\"].items()}\n train_spots = {k: np.array(v) for k, v in h5file[\"spots/Train\"].items()}\n test_images = {k: np.array(v) for k, v in h5file[\"images/Test\"].items()}\n test_spots = {k: np.array(v) for k, v in h5file[\"spots/Test\"].items()}\n return train_images, train_spots, test_images, test_spots","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:04:19.824395Z","iopub.execute_input":"2025-03-19T10:04:19.824937Z","iopub.status.idle":"2025-03-19T10:04:19.831584Z","shell.execute_reply.started":"2025-03-19T10:04:19.824902Z","shell.execute_reply":"2025-03-19T10:04:19.830378Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"file_path = \"/kaggle/input/el-hackathon-2025/elucidata_ai_challenge_data.h5\" \ntrain_images, train_spots, test_images, test_spots = load_data(file_path)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:04:35.396666Z","iopub.execute_input":"2025-03-19T10:04:35.397066Z","iopub.status.idle":"2025-03-19T10:04:37.483664Z","shell.execute_reply.started":"2025-03-19T10:04:35.397034Z","shell.execute_reply":"2025-03-19T10:04:37.482595Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"print(\"Dataset Overview:\")\nprint(f\" Number of training slides: {len(train_images)}\")\nprint(f\" Number of test slides: {len(test_images)}\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:04:48.001060Z","iopub.execute_input":"2025-03-19T10:04:48.001408Z","iopub.status.idle":"2025-03-19T10:04:48.007966Z","shell.execute_reply.started":"2025-03-19T10:04:48.001382Z","shell.execute_reply":"2025-03-19T10:04:48.006861Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"print(\"\\nTraining Images Structure:\")\nfor slide_name, image in train_images.items():\n print(f\" Slide: {slide_name}, Shape: {image.shape}, Data Type: {image.dtype}\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:09:36.722449Z","iopub.execute_input":"2025-03-19T10:09:36.722829Z","iopub.status.idle":"2025-03-19T10:09:36.729441Z","shell.execute_reply.started":"2025-03-19T10:09:36.722795Z","shell.execute_reply":"2025-03-19T10:09:36.728498Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"print(\"\\nTraining Spots Structure:\")\nfor slide_name, spots in train_spots.items():\n print(f\" Slide: {slide_name}, Number of Spots: {len(spots)}, Data Type: {spots.dtype}\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:09:39.570807Z","iopub.execute_input":"2025-03-19T10:09:39.571462Z","iopub.status.idle":"2025-03-19T10:09:39.579668Z","shell.execute_reply.started":"2025-03-19T10:09:39.571399Z","shell.execute_reply":"2025-03-19T10:09:39.578585Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"print(\"\\nTest Images Structure:\")\nfor slide_name, image in test_images.items():\n print(f\" Slide: {slide_name}, Shape: {image.shape}, Data Type: {image.dtype}\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:09:45.647775Z","iopub.execute_input":"2025-03-19T10:09:45.648095Z","iopub.status.idle":"2025-03-19T10:09:45.653459Z","shell.execute_reply.started":"2025-03-19T10:09:45.648061Z","shell.execute_reply":"2025-03-19T10:09:45.652549Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"print(\"\\nTest Spots Structure:\")\nfor slide_name, spots in test_spots.items():\n print(f\" Slide: {slide_name}, Number of Spots: {len(spots)}, Data Type: {spots.dtype}\")","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:09:48.199989Z","iopub.execute_input":"2025-03-19T10:09:48.200341Z","iopub.status.idle":"2025-03-19T10:09:48.206083Z","shell.execute_reply.started":"2025-03-19T10:09:48.200301Z","shell.execute_reply":"2025-03-19T10:09:48.205120Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"# Analysis of the Output\n\n## 📌 Image Shapes:\n- The **training images** have slightly varying dimensions, all around **2000 pixels** in one dimension and just under that in the other, with **3 color channels (RGB)**.\n- The **test image** is also around **2000x2000 pixels**.\n- This variation in size will be important to consider during preprocessing (e.g., **resizing or padding**).\n\n## 📌 Number of Spots:\n- The number of spots **varies significantly** between training slides, ranging from **328 to 2270**.\n- This indicates **varying tissue areas or spot densities**.\n- The **test slide** has **2088 spots**.\n\n## 📌 Spot Data Structure:\n- Training spot data includes:\n - **x, y coordinates**.\n - **35 cell type abundances** (**C1 to C35**).","metadata":{}},{"cell_type":"markdown","source":"## Image Size Analysis: Training and Test Sets","metadata":{}},{"cell_type":"code","source":"# Image Size Distribution\ntrain_widths = [image.shape[1] for image in train_images.values()]\ntrain_heights = [image.shape[0] for image in train_images.values()]\ntest_widths = [image.shape[1] for image in test_images.values()]\ntest_heights = [image.shape[0] for image in test_images.values()]","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:17:44.738612Z","iopub.execute_input":"2025-03-19T10:17:44.738984Z","iopub.status.idle":"2025-03-19T10:17:44.744521Z","shell.execute_reply.started":"2025-03-19T10:17:44.738954Z","shell.execute_reply":"2025-03-19T10:17:44.743242Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"sns.set_style(\"whitegrid\")\nsns.set_palette(\"coolwarm\")\n\nfig, axes = plt.subplots(1, 2, figsize=(14, 6), dpi=100)\n\nsns.histplot(train_widths, kde=True, bins=30, ax=axes[0], color=\"#4c72b0\", edgecolor=\"black\", alpha=0.8)\naxes[0].set_title(\"Train Image Widths Distribution\", fontsize=14, fontweight='bold', color=\"#333\")\naxes[0].set_xlabel(\"Width\", fontsize=12)\naxes[0].set_ylabel(\"Frequency\", fontsize=12)\n\nsns.histplot(train_heights, kde=True, bins=30, ax=axes[1], color=\"#dd8452\", edgecolor=\"black\", alpha=0.8)\naxes[1].set_title(\"Train Image Heights Distribution\", fontsize=14, fontweight='bold', color=\"#333\")\naxes[1].set_xlabel(\"Height\", fontsize=12)\naxes[1].set_ylabel(\"Frequency\", fontsize=12)\n\nplt.tight_layout()\nplt.show()\n\n# Test Image Width Distribution\nplt.figure(figsize=(7, 4), dpi=100)\nsns.histplot(test_widths, kde=True, bins=30, color=\"#55a868\", edgecolor=\"black\", alpha=0.8)\nplt.title(\"Test Image Widths Distribution\", fontsize=14, fontweight='bold', color=\"#333\")\nplt.xlabel(\"Width\", fontsize=12)\nplt.ylabel(\"Frequency\", fontsize=12)\nplt.grid(alpha=0.3)\nplt.show()\n\n# Test Image Height Distribution\nplt.figure(figsize=(7, 4), dpi=100)\nsns.histplot(test_heights, kde=True, bins=30, color=\"#c44e52\", edgecolor=\"black\", alpha=0.8)\nplt.title(\"Test Image Heights Distribution\", fontsize=14, fontweight='bold', color=\"#333\")\nplt.xlabel(\"Height\", fontsize=12)\nplt.ylabel(\"Frequency\", fontsize=12)\nplt.grid(alpha=0.3)\nplt.show()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:18:49.511672Z","iopub.execute_input":"2025-03-19T10:18:49.512247Z","iopub.status.idle":"2025-03-19T10:18:51.024621Z","shell.execute_reply.started":"2025-03-19T10:18:49.512215Z","shell.execute_reply":"2025-03-19T10:18:51.023682Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
📊 Image Size Analysis: Training and Test Sets
\n\n
🖼️ Training Images:
\n\n\n
🖼️ Test Images:
\n","metadata":{}},{"cell_type":"markdown","source":"## Color Channel Analysis","metadata":{}},{"cell_type":"code","source":"def analyze_color_channels(images):\n channel_means = {}\n channel_stds = {}\n for slide_name, image in images.items():\n # Flatten the image to calculate statistics across all pixels\n pixels = image.reshape(-1, 3)\n channel_means[slide_name] = np.mean(pixels, axis=0)\n channel_stds[slide_name] = np.std(pixels, axis=0)\n return channel_means, channel_stds","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:20:17.718379Z","iopub.execute_input":"2025-03-19T10:20:17.718707Z","iopub.status.idle":"2025-03-19T10:20:17.724305Z","shell.execute_reply.started":"2025-03-19T10:20:17.718681Z","shell.execute_reply":"2025-03-19T10:20:17.723165Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"train_channel_means, train_channel_stds = analyze_color_channels(train_images)\ntest_channel_means, test_channel_stds = analyze_color_channels(test_images)","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:20:26.554750Z","iopub.execute_input":"2025-03-19T10:20:26.555123Z","iopub.status.idle":"2025-03-19T10:20:28.333374Z","shell.execute_reply.started":"2025-03-19T10:20:26.555087Z","shell.execute_reply":"2025-03-19T10:20:28.332265Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"sns.set_style(\"whitegrid\")\n\nnum_slides = len(train_channel_means)\ncols = 3 \nrows = (num_slides // cols) + (num_slides % cols > 0) \n\nfig, axes = plt.subplots(rows, cols, figsize=(14, 5 * rows), dpi=100)\n\naxes = axes.flatten() if num_slides > 1 else [axes]\n\ncolors = [\"#e63946\", \"#2a9d8f\", \"#457b9d\"] \n\nfor i, (slide_name, means) in enumerate(train_channel_means.items()):\n axes[i].bar([\"Red\", \"Green\", \"Blue\"], means, color=colors, edgecolor=\"black\", alpha=0.85)\n axes[i].set_title(f\"Train {slide_name} - Channel Means\", fontsize=14, fontweight=\"bold\", color=\"#333\")\n axes[i].set_ylabel(\"Mean Pixel Value\", fontsize=12)\n axes[i].set_ylim(0, 1) \n axes[i].set_xticklabels([\"Red\", \"Green\", \"Blue\"], fontsize=12)\n\nplt.tight_layout()\nplt.show()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:21:26.618933Z","iopub.execute_input":"2025-03-19T10:21:26.619277Z","iopub.status.idle":"2025-03-19T10:21:27.807392Z","shell.execute_reply.started":"2025-03-19T10:21:26.619251Z","shell.execute_reply":"2025-03-19T10:21:27.805999Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
Channel Means Interpretation (Training Images)
\n\n","metadata":{}},{"cell_type":"code","source":"sns.set_style(\"whitegrid\")\n\nnum_slides = len(train_channel_stds)\ncols = 3 \nrows = (num_slides // cols) + (num_slides % cols > 0) \n\nfig, axes = plt.subplots(rows, cols, figsize=(17, 4 * rows), dpi=100)\n\naxes = axes.flatten() if num_slides > 1 else [axes]\n\ncolors = [\"#e63946\", \"#2a9d8f\", \"#457b9d\"]\n\nfor i, (slide_name, stds) in enumerate(train_channel_stds.items()):\n axes[i].bar([\"Red\", \"Green\", \"Blue\"], stds, color=colors, edgecolor=\"black\", alpha=0.85)\n axes[i].set_title(f\"Train {slide_name} - Channel Stds\", fontsize=14, fontweight=\"bold\", color=\"#333\")\n axes[i].set_ylabel(\"Standard Deviation\", fontsize=12)\n axes[i].set_ylim(0, 1) \n axes[i].set_xticklabels([\"Red\", \"Green\", \"Blue\"], fontsize=12)\n\n\nplt.tight_layout()\nplt.show()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:23:08.881587Z","iopub.execute_input":"2025-03-19T10:23:08.881957Z","iopub.status.idle":"2025-03-19T10:23:10.189816Z","shell.execute_reply.started":"2025-03-19T10:23:08.881930Z","shell.execute_reply":"2025-03-19T10:23:10.188807Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
Channel Standard Deviations Interpretation (Training Images)
\n\n","metadata":{}},{"cell_type":"markdown","source":"## Spots per Image","metadata":{}},{"cell_type":"code","source":"spot_counts = {\n slide_name: len(spots) for slide_name, spots in train_spots.items()\n}\n\nplt.figure(figsize=(8, 5))\nsns.barplot(x=list(spot_counts.keys()), y=list(spot_counts.values()))\nplt.title(\"Number of Spots per Training Image\")\nplt.xlabel(\"Slide\")\nplt.ylabel(\"Number of Spots\")\nplt.show()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:34:14.724687Z","iopub.execute_input":"2025-03-19T10:34:14.725113Z","iopub.status.idle":"2025-03-19T10:34:14.961545Z","shell.execute_reply.started":"2025-03-19T10:34:14.725083Z","shell.execute_reply":"2025-03-19T10:34:14.960522Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"sns.set_style(\"whitegrid\")\n\nnum_slides = len(train_spots)\ncols = 3 \nrows = (num_slides // cols) + (num_slides % cols > 0) \n\nfig, axes = plt.subplots(rows, cols, figsize=(16, 6 * rows), dpi=100)\n\naxes = axes.flatten() if num_slides > 1 else [axes]\n\n# Aesthetic color palette\nspot_color = \"#000000\" # black color for spots\n\nfor i, (slide_name, spots) in enumerate(train_spots.items()):\n axes[i].scatter(\n spots[\"x\"], spots[\"y\"], s=5, color=spot_color, alpha=1, edgecolors='w', linewidth=0.5\n ) \n axes[i].set_title(f\"Spot Distribution - {slide_name}\", fontsize=14, fontweight=\"bold\", color=\"#ff0808\")\n axes[i].set_xlabel(\"X Coordinate\", fontsize=12)\n axes[i].set_ylabel(\"Y Coordinate\", fontsize=12)\n axes[i].invert_yaxis() \n axes[i].tick_params(axis='both', labelsize=10) \n\nplt.tight_layout()\nplt.show()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:39:17.074284Z","iopub.execute_input":"2025-03-19T10:39:17.074614Z","iopub.status.idle":"2025-03-19T10:39:19.147916Z","shell.execute_reply.started":"2025-03-19T10:39:17.074588Z","shell.execute_reply":"2025-03-19T10:39:19.146840Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
🔍 Spatial Distribution Analysis
\n\n
\n 📊 Varying Spot Density:\n \n
\n\n
\n 🔢 Shape and Coverage:\n \n
\n\n
\n ⚠️ Edge Effects:\n \n
\n\n
\n 🔍 Slide S_3 and S_6:\n \n
","metadata":{}},{"cell_type":"markdown","source":"## Cell Type Analysis - Abundance Distribution","metadata":{}},{"cell_type":"code","source":"train_spot_dfs = {\n slide_name: pd.DataFrame(spots) for slide_name, spots in train_spots.items()\n}\n\nall_train_spots_df = pd.concat(train_spot_dfs.values(), ignore_index=True)\n\ncell_type_cols = [\n col for col in all_train_spots_df.columns if col.startswith(\"C\")\n] \ncell_type_stats = all_train_spots_df[cell_type_cols].describe()\n\nprint(\"Descriptive Statistics for Cell Type Abundances:\")\ncell_type_stats","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:49:38.441763Z","iopub.execute_input":"2025-03-19T10:49:38.442169Z","iopub.status.idle":"2025-03-19T10:49:38.556928Z","shell.execute_reply.started":"2025-03-19T10:49:38.442138Z","shell.execute_reply":"2025-03-19T10:49:38.556038Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"
📊 Analysis of Cell Type Abundance Statistics
\n\n
\n 🔢 Varying Abundance Ranges:\n \n
\n\n
\n 📉 High Standard Deviations:\n \n
\n\n
\n 📊 Skewed Distributions:\n \n
\n\n
\n 🔽 Minimum Values:\n \n
\n\n
📈 Implications for Modeling
\n\n
\n 🔄 Data Transformation:\n \n
\n\n
\n ⚖️ Class Imbalance:\n \n
\n\n
\n 📉 Outlier Sensitivity:\n \n
","metadata":{}},{"cell_type":"markdown","source":"## Correlation Matrix of Cell Type Abundances","metadata":{}},{"cell_type":"code","source":"cell_type_corr = all_train_spots_df[cell_type_cols].corr()\n\nplt.figure(figsize=(20, 12))\nsns.heatmap(cell_type_corr, annot=False, cmap=\"Blues\") \nplt.title(\"Correlation Matrix of Cell Type Abundances\")\nplt.show()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T10:59:03.604438Z","iopub.execute_input":"2025-03-19T10:59:03.604839Z","iopub.status.idle":"2025-03-19T10:59:04.564205Z","shell.execute_reply.started":"2025-03-19T10:59:03.604807Z","shell.execute_reply":"2025-03-19T10:59:04.563071Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"# Analysis of Cell Type Correlations\n\n### Positive Correlations:\n* There are several areas of lighter blue, indicating positive correlations between certain cell types. This suggests that these cell types tend to co-occur or have similar abundance patterns.\n* Some areas of positive correlation can be seen among C1, C5, and C27.\n\n### Negative Correlations:\n* There are some areas of very light colors, almost white, indicating potential negative correlations between certain cell types. This suggests that when one cell type has a high abundance, the other tends to have a low abundance.\n\n### Clustering Patterns:\n* The heatmap reveals some clustering patterns, where groups of cell types exhibit similar correlation profiles. This might indicate that these cell types are functionally related or belong to the same biological pathways.\n\n### Sparse Correlations:\n* Most of the heatmap is a light blue, this means that most of the cell types do not have a strong correlation with other cell types.\n\n# Implications for Modeling\n\n### Feature Engineering:\n* The identified correlations can be used to inform feature engineering. For example, we might create new features that represent the ratios or differences between highly correlated cell types.\n\n### Model Selection:\n* Models that can capture complex relationships and interactions between features, such as tree-based models or neural networks, might be more suitable for this data.\n\n### Dimensionality Reduction:\n* If some cell types are highly correlated, we might consider dimensionality reduction techniques (e.g., PCA) to reduce the number of input features and improve model efficiency.\n\n### Multicollinearity:\n* The positive correlations that are present, mean that multicollinearity is also present, and that must be kept in mind when building the model.","metadata":{}},{"cell_type":"markdown","source":"## Image Visualization","metadata":{}},{"cell_type":"code","source":"plt.figure(figsize=(12, 6))\nfor i, slide_name in enumerate(list(train_images.keys())[:3]):\n plt.subplot(1, 3, i + 1)\n plt.imshow(train_images[slide_name])\n plt.title(f\"Train - {slide_name}\")\n plt.axis(\"off\") \nplt.tight_layout()\nplt.show()\n\nplt.figure(figsize=(6, 6))\n# Display the test image\nplt.imshow(test_images[\"S_7\"])\nplt.title(\"Test - S_7\")\nplt.axis(\"off\")\nplt.show()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T11:16:15.277355Z","iopub.execute_input":"2025-03-19T11:16:15.277696Z","iopub.status.idle":"2025-03-19T11:16:19.387425Z","shell.execute_reply.started":"2025-03-19T11:16:15.277670Z","shell.execute_reply":"2025-03-19T11:16:19.386237Z"}},"outputs":[],"execution_count":null},{"cell_type":"code","source":"test_spots_df = pd.DataFrame(test_spots[\"S_7\"])\n\ntest_set_counts = test_spots_df[\"Test_Set\"].value_counts()\n\nprint(\"Value Counts for 'Test_Set' Column:\")\nprint(test_set_counts)\n\nplt.figure(figsize=(6, 4))\nsns.barplot(x=test_set_counts.index, y=test_set_counts.values)\nplt.title(\"Distribution of Test_Set Values\")\nplt.xlabel(\"Test_Set Value\")\nplt.ylabel(\"Count\")\nplt.show()","metadata":{"trusted":true,"execution":{"iopub.status.busy":"2025-03-19T11:17:22.958651Z","iopub.execute_input":"2025-03-19T11:17:22.959067Z","iopub.status.idle":"2025-03-19T11:17:23.167285Z","shell.execute_reply.started":"2025-03-19T11:17:22.959038Z","shell.execute_reply":"2025-03-19T11:17:23.166190Z"}},"outputs":[],"execution_count":null},{"cell_type":"markdown","source":"## Analysis of the Test_Set Column\n\n### **Unequal Distribution:** The values are not evenly distributed. Value 2 appears significantly more frequently (1586 times) than value 1 (502 times).\n","metadata":{}},{"cell_type":"markdown","source":"
\n

Ready for Model Building?

\n

\n We've completed a thorough Exploratory Data Analysis of the histology image data. We've found some key insights into image properties, cell type distributions, and spatial relationships, etc\n

\n

\n Now, it's time to use these insights to build predictive models! Stay tuned for our upcoming notebook where we will implement various machine learning techniques to tackle the cell type prediction challenge.\n

\n
\n

\n \n STAY TUNED! (Coming Soon)\n \n

\n

\n If you found this EDA notebook helpful, please consider giving it an upvote! Your support is greatly appreciated.\n

\n
\n
","metadata":{}}]}