{"id":12630,"date":"2021-12-23T08:01:18","date_gmt":"2021-12-23T16:01:18","guid":{"rendered":"https:\/\/www.couchbase.com\/blog\/?p=12630"},"modified":"2024-08-30T03:03:43","modified_gmt":"2024-08-30T10:03:43","slug":"how-couchbase-simplifies-data-science-part-2","status":"publish","type":"post","link":"https:\/\/www.couchbase.com\/blog\/ko\/how-couchbase-simplifies-data-science-part-2\/","title":{"rendered":"\uce74\uc6b0\uce58\ubca0\uc774\uc2a4\uac00 \ub370\uc774\ud130 \uacfc\ud559\uc744 \uac04\uc18c\ud654\ud558\ub294 \ubc29\ubc95(2\ubd80)"},"content":{"rendered":"

\uc774 \ube14\ub85c\uadf8\ub294 \uace0\ub4f1\ud559\uc0dd \uc778\ud134\uc778 Karen Yuan\uc774 \uacf5\ub3d9 \uc791\uc131\ud588\uc2b5\ub2c8\ub2e4.<\/em><\/p>\n

\uc6b0\ub9ac\uc758 \uc774\uc804 \uae00<\/a> Couchbase \ucffc\ub9ac \uc11c\ube44\uc2a4\ub97c \uc0ac\uc6a9\ud558\uc5ec \ud0d0\uc0c9\uc801 \ub370\uc774\ud130 \ubd84\uc11d\uc744 \uc218\ud589\ud558\ub294 \ubc29\ubc95\uc744 \ubc30\uc6e0\uc2b5\ub2c8\ub2e4. \ub610\ud55c Couchbase Python SDK\uc758 \ucffc\ub9ac API\ub97c \uc0ac\uc6a9\ud558\uc5ec \ud559\uc2b5 \ub370\uc774\ud130\ub97c \ud6a8\uc728\uc801\uc73c\ub85c \uc77d\uace0 \uba38\uc2e0 \ub7ec\ub2dd(ML)\uc5d0 \uc801\ud569\ud55c \ud310\ub2e4\uc2a4 \ub370\uc774\ud130 \ud504\ub808\uc784\uc5d0 \uc6d0\ud65c\ud558\uac8c \uc800\uc7a5\ud558\ub294 \ubc29\ubc95\ub3c4 \ubc30\uc6e0\uc2b5\ub2c8\ub2e4. \uadf8\ub9ac\uace0 \ub9c8\uc9c0\ub9c9\uc73c\ub85c ML \ubaa8\ub378\uacfc \ud574\ub2f9 \uba54\ud0c0\ub370\uc774\ud130\ub97c Couchbase\uc5d0 \uc800\uc7a5\ud588\uc2b5\ub2c8\ub2e4. \uc774 \uae00\uc5d0\uc11c\ub294 \uc608\uce21\uc744 \uc0dd\uc131\ud558\uace0, \uc774\ub97c Couchbase\uc5d0 \uc800\uc7a5\ud558\uace0, \ucffc\ub9ac \ucc28\ud2b8\ub97c \uc0ac\uc6a9\ud558\uc5ec \ubd84\uc11d\ud558\ub294 \ubc29\ubc95\uc5d0 \ub300\ud574 \uc54c\uc544\ubcf4\uaca0\uc2b5\ub2c8\ub2e4.<\/p>\n

\uc2e4\uc2dc\uac04 \uc608\uce21<\/h4>\n

\ub370\uc774\ud130 \uacfc\ud559\uc790\ub294 \ud559\uc2b5\ub41c \ubaa8\ub378\uc744 \uc0ac\uc6a9\ud558\uc5ec \uc608\uce21\uc744 \uc0dd\uc131\ud569\ub2c8\ub2e4.<\/p>\n

\uadf8\ub9bc 1\uc758 \uc608\uce21 \ud750\ub984\uc744 \uc0ac\uc6a9\ud558\uc5ec \uc2e4\uc2dc\uac04\uc73c\ub85c \uc774\ud0c8 \uc810\uc218\ub97c \uc608\uce21\ud558\uace0 \uc608\uce21 \uacb0\uacfc\ub97c Couchbase\uc5d0 \uc800\uc7a5\ud558\uaca0\uc2b5\ub2c8\ub2e4. \uc774\uc804 \uae00\uc5d0\uc11c \ud559\uc2b5\ud55c \uc774\ud0c8 \uc608\uce21 \ubaa8\ub378\uc744 \uc0ac\uc6a9\ud558\uaca0\uc2b5\ub2c8\ub2e4.<\/p>\n

\"Real<\/p>\n

\uce74\uc6b0\uce58\ubca0\uc774\uc2a4\uc5d0 \uc800\uc7a5\ub41c \ubaa8\ub378\uacfc \ud574\ub2f9 \uba54\ud0c0\ub370\uc774\ud130\ub97c \uc77d\ub294 \ud568\uc218\uc785\ub2c8\ub2e4:<\/strong><\/p>\n

def read_model_from_couchbase(model_id):\r\n\u00a0 \u00a0 \u00a0bucket = cluster.bucket('model_repository')\r\n\u00a0 \u00a0 \u00a0model_bytes = bucket.get(model_id).value\r\n\u00a0 \u00a0 \u00a0model = pickle.loads(model_bytes)\r\n\u00a0 \u00a0 \u00a0key = model_id + \"_metadata\"\r\n\u00a0 \u00a0 \u00a0feature_names = bucket.get(key).value['feature_names']\r\n\u00a0 \u00a0 \u00a0return {'model': model, 'feature_names': set(feature_names)}<\/pre>\n
\uce74\uc6b0\uce58\ubca0\uc774\uc2a4\uc5d0 \uc800\uc7a5\ub41c \uace0\uac1d \ub370\uc774\ud130\ub97c \uc77d\ub294 \uae30\ub2a5\uc785\ub2c8\ub2e4:<\/strong><\/p>\n
# We will use the Query interface in the Couchbase Python SDK to get multiple customer\r\n# records that satisfy a condition. Alternatively, the GET interface in the Couchbase Python SDK can \r\n# be used to get individual customer records e.g. bucket.get(customer_key).value\r\ndef read_data_from_couchbase(select_clause = \"\", where_clause = \"\"):\r\n     if select_clause:\r\n        query_statement = \"SELECT customer.*, \" + select_clause \r\n     else:\r\n        query_statement = \"SELECT customer.*\"\r\n\r\n     query_statement = query_statement + \" FROM `online_streaming` as customer\"\r\n\r\n     if where_clause:\r\n        query_statement = query_statement + \" WHERE \" + where_clause\r\n     # Use the Query API to get customer records\r\n     query_result = cb.query(query_statement)\r\n     return pd.DataFrame.from_dict(list(query_result))<\/pre>\n
\ub2e4\uc74c \uc0ac\ud56d \uc608\uce21<\/em> \ud568\uc218\ub294 \uc704\uc758 \ud568\uc218\ub97c \uc0ac\uc6a9\ud558\uc5ec \ubaa8\ub378, \uba54\ud0c0\ub370\uc774\ud130 \ubc0f \uace0\uac1d \ub808\ucf54\ub4dc\ub97c \uc77d\uc2b5\ub2c8\ub2e4. \ud559\uc2b5 \uc911\uc5d0 \uc0ac\uc6a9\ub41c \uac83\uacfc \ub3d9\uc77c\ud55c \ud504\ub85c\uc138\uc2a4(\uc989, \uc6d0\ud56b \uc778\ucf54\ub529)\ub97c \uc0ac\uc6a9\ud558\uc5ec \uace0\uac1d \ub370\uc774\ud130\ub97c \ud53c\ucc98\ub85c \ubcc0\ud658\ud569\ub2c8\ub2e4. \uadf8\ub7f0 \ub2e4\uc74c \uc774 \ud2b9\uc9d5\uc5d0 \ub300\ud574 \ubaa8\ub378\uc744 \uc2e4\ud589\ud558\uc5ec \uc774\ud0c8 \uc810\uc218\ub97c \uc608\uce21\ud569\ub2c8\ub2e4.<\/p>\n
def predict(model_id, select_clause = \"\", where_clause = \"\"):\r\n     # Step 1: Read the model and its metadata from Couchbase\r\n     rv = read_model_from_couchbase(model_id)\r\n     model = rv['model']\r\n     feature_names = rv['feature_names']\r\n\r\n     # Step 2: Read customer records from Couchbase\r\n     df = read_data_from_couchbase(select_clause, where_clause)\r\n     customer_prediction = df\r\n\r\n     # Step 3: Convert the raw data into features expected by the model\r\n     df = pd.get_dummies(df)\r\n     drop_cols = set(list(df.columns)) - feature_names\r\n     df.drop(drop_cols, axis = 1, inplace = True)\r\n     df = df.reindex(columns=feature_names, fill_value=0)\r\n     # Step 4: Predict\r\n     prediction = model.predict(df)\r\n     customer_prediction['Churn Prediction'] = prediction\r\n     return customer_prediction\r\n<\/pre>\n
# Example: Predict churn for a customer with given ID \r\nprediction = predict('churn_predictor_model_v1', where_clause = \"customer.CustomerID = 100002\")[['CustomerID', 'Churn Prediction']] \r\nprediction<\/pre>\n
\ub2e4\uc74c \ub300\uc0c1\uc5d0 \ub300\ud55c \uc774\ud0c8 \uc608\uce21 customerID 100002<\/em> \uc774 1\uc774\uba74 \uc2a4\ud2b8\ub9ac\ubc0d \uc11c\ube44\uc2a4\ub97c \ub5a0\ub0a0 \uac00\ub2a5\uc131\uc774 \ub192\ub2e4\ub294 \ub73b\uc785\ub2c8\ub2e4.<\/p>\n
\"churn<\/p>\n
\uc608\uce21\uc740 \uc544\ub798 \ud45c\uc2dc\ub41c \ucf54\ub4dc\ub97c \uc0ac\uc6a9\ud558\uc5ec \uc608\uce21\uc774\ub77c\ub294 Couchbase \ubc84\ud0b7\uc5d0 \uc800\uc7a5\ub429\ub2c8\ub2e4. \uacc4\uc18d \uc9c4\ud589\ud558\uae30 \uc804\uc5d0 Couchbase \ud074\ub7ec\uc2a4\ud130\uc5d0 \uc608\uce21 \ubc84\ud0b7\uc744 \ub9cc\ub4dc\uc138\uc694.<\/p>\n
bucket = cluster.bucket('predictions')\r\nto_save = prediction.to_dict(orient=\"records\")[0]\r\nkey = str(prediction.iloc[0]['CustomerID'])\r\nbucket.upsert(key, to_save)<\/pre>\n
\uc608\uce21\uc774 Couchbase\uc5d0 \uc131\uacf5\uc801\uc73c\ub85c \uc800\uc7a5\ub418\uc5c8\ub294\uc9c0 \ud655\uc778\ud569\ub2c8\ub2e4.<\/p>\n
\"\"<\/p>\n
Python UDF \uae30\ub2a5(\ud604\uc7ac \uac1c\ubc1c\uc790 \ud504\ub9ac\ubdf0 \ubc84\uc804)\uc744 \uc0ac\uc6a9\ud558\uc5ec \ud559\uc2b5\ub41c \ubaa8\ub378\uc744 \uc2e4\ud589\ud558\uace0 Couchbase Analytics\uc5d0\uc11c \uc608\uce21\uc744 \uc0dd\uc131\ud560 \uc218\ub3c4 \uc788\uc2b5\ub2c8\ub2e4. \ub2e4\uc74c \ubb38\uc11c\ub97c \ucc38\uc870\ud558\uc138\uc694. \uce74\uc6b0\uce58\ubca0\uc774\uc2a4 \uc560\ub110\ub9ac\ud2f1\uc2a4 \ud30c\uc774\uc36c UDF\ub97c \uc0ac\uc6a9\ud558\uc5ec ML \ubaa8\ub378 \uc2e4\ud589\ud558\uae30<\/a> \uc5d0\uc11c \uc790\uc138\ud55c \ub0b4\uc6a9\uc744 \ud655\uc778\ud558\uc138\uc694.<\/p>\n
\uac00\uc815 \ubd84\uc11d<\/h4>\n
\ub370\uc774\ud130 \uacfc\ud559\uc790\ub294 \uc608\uce21\uc744 \ubd84\uc11d\ud558\uc5ec \uc758\uc0ac \uacb0\uc815\uc5d0 \ub3c4\uc6c0\uc774 \ub418\ub294 \uc9c8\ubb38\uc5d0 \ub2f5\ud569\ub2c8\ub2e4.<\/p>\n
\uc774\uc804 \uae00\uc5d0\uc11c \uc815\uc758\ud55c \ubb38\uc81c\ub294 \uc628\ub77c\uc778 \uc2a4\ud2b8\ub9ac\ubc0d \uc11c\ube44\uc2a4 \ud68c\uc0ac\uc758 \uc601\uc5c5\ud300\uc774 \uc6d4\ubcc4 \ube44\uc6a9\uc744 \ub298\ub9ac\uba74 \uace0\uac1d \uc774\ud0c8\uc744 \uc5b5\uc81c\ud558\uba74\uc11c \uc218\uc775\uc744 \uadf9\ub300\ud654\ud560 \uc218 \uc788\ub294\uc9c0 \uc54c\uace0 \uc2f6\uc5b4\ud558\ub294 \uac83\uc774\uc5c8\uc2b5\ub2c8\ub2e4.<\/p>\n
\uc774\uc5d0 \ub300\ud55c \ub2f5\uc744 \uad6c\ud558\uae30 \uc704\ud574 \uc544\ub798 \ucf54\ub4dc\ub97c \uc0ac\uc6a9\ud558\uc5ec \uc6d4\ubcc4 \ube44\uc6a9\uc774 $1, $2 \ub4f1\uc73c\ub85c \uc99d\uac00\ud588\uc744 \ub54c\uc758 \uc774\ud0c8 \uc810\uc218\ub97c \uc608\uce21\ud569\ub2c8\ub2e4. \uc774 \ubd84\uc11d \uacb0\uacfc\ub294 \uc608\uce21 \ubc84\ud0b7\uc5d0 \uc800\uc7a5\ub429\ub2c8\ub2e4.
\nCouchbase \ud074\ub7ec\uc2a4\ud130 UI\ub97c \uc0ac\uc6a9\ud558\uc5ec \ub2e4\uc74c\uacfc \uac19\uc740 \ubc94\uc704\ub97c \ub9cc\ub4ed\ub2c8\ub2e4. what_if_analysis<\/em> \ub77c\ub294 \uceec\ub809\uc158\uacfc \uc99d\uac00_\uc6d4\ubcc4_\ube44\uc6a9<\/em> \uc744 \uc608\uce21 \ubc84\ud0b7\uc5d0 \ucd94\uac00\ud569\ub2c8\ub2e4. (\ubc94\uc704 \ubc0f \uceec\ub809\uc158\uc740 Couchbase Server 7.0 \uc774\uc0c1\uc5d0\uc11c \uc0ac\uc6a9\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.)<\/p>\n
# Connect to predictions bucket on Couchbase and relevant scope and collection\r\nbucket = cluster.bucket('predictions')\r\nwa_scope = bucket.scope(\"what_if_analysis\")\r\nwa_collection = wa_scope.collection(\"increase_monthly_cost\")\r\n# Predict the churn rate if the monthly cost of existing customers is increased by \r\n# $1, $2 .. $5\r\nfor increase_by in range (1, 6):\r\n      # where_cluase is set to customer.Churn = 0 because we are interested \r\n      # only in existing customers\r\n      # The following select_clause will update the MonthlyCost as shown while returning the\r\n      # query result. In each iteration of this loop the MonthlyCost will increase by $1, $2 ...\r\n      select_clause = \"customer.`MonthlyCost` + \" + str(increase_by) + \" as `MonthlyCost`\"\r\n      # The updated MonthlyCost is passed as input along with other attributes to the prediction\r\n      # function. The output tells which of the existing customers are likely to\r\n      # to churn if the MonthlyCost is increased by the specified amount. \r\n      rv = predict('churn_predictor_model_v1', select_clause = select_clause, where_clause = \"customer.Churn = 0\")\r\n      # Monthly revenues are predicted by adding the monthly cost of the customers not likely to churn.\r\n      rv['Predicted Monthly Revenue'] = (1 - rv['Churn Prediction']) * rv['MonthlyCost']\r\n      predicted_churn_rate = rv['Churn Prediction'].value_counts(normalize=True).mul(100)[1]\r\n      # Save predictions to Couchbase.\r\n      to_save = {'Monthly Cost Increase ($)': increase_by, \r\n      'Predicted Monthly Revenue': rv['Predicted Monthly Revenue'].sum(),\r\n      'Predicted Churn Rate': predicted_churn_rate}\r\n       key = \"increase_by_$\" + str(increase_by)\r\n       wa_collection.upsert(key, to_save)\r\n\r\n# Use the Query API to calculate current revenue. Store it also on the predictions bucket. This is \r\n# used for comparison\r\ncurrent = cb.query('SELECT sum(customer.`MonthlyCost`) as CurrentRevenue FROM `online_streaming` customer where customer.Churn = 0')\r\nto_save = {'Monthly Cost Increase ($)': 0, 'Predicted Monthly Revenue': list(current)[0]['CurrentRevenue'], 'Predicted Churn Rate': 0}\r\nwa_collection.upsert(\"current\", to_save)<\/pre>\n
\uce74\uc6b0\uce58\ubca0\uc774\uc2a4 \ucffc\ub9ac\ub97c \uc0ac\uc6a9\ud558\uc5ec \uc608\uce21 \uacb0\uacfc\ub97c \ubd84\uc11d\ud558\ub824\uba74 \ub2e4\uc74c\uacfc \uac19\uc774 \uae30\ubcf8 \uc778\ub371\uc2a4\ub97c \ub9cc\ub4ed\ub2c8\ub2e4. what_if_analysis<\/em> \ubc94\uc704\ub97c \uc124\uc815\ud569\ub2c8\ub2e4. \ucffc\ub9ac \ucee8\ud14d\uc2a4\ud2b8\ub294 \uadf8\ub9bc\uacfc \uac19\uc774 \uc124\uc815\ud574\uc57c \ud569\ub2c8\ub2e4.<\/p>\n
\"query
\n\ucffc\ub9ac \ucc28\ud2b8\ub97c \uc0ac\uc6a9\ud558\uc5ec \uc608\uce21 \uacb0\uacfc\ub97c \ubd84\uc11d\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uc544\ub798 \ucc28\ud2b8\uc5d0\uc11c\ub294 \uc6d4 \ube44\uc6a9\uc774 $1 \uc99d\uac00\ud560 \uacbd\uc6b0 \uae30\uc874 \uace0\uac1d \uc911 \uc57d 7%\uac00 \uc774\ud0c8\ud560 \uac83\uc73c\ub85c \uc608\uce21\ub418\uace0, \uc6d4 \ube44\uc6a9\uc774 $2 \uc99d\uac00\ud560 \uacbd\uc6b0 \uc57d 10%\uac00 \uc774\ud0c8\ud560 \uac83\uc73c\ub85c \uc608\uce21\ub418\ub294 \ub4f1 \uae30\uc874 \uace0\uac1d\uc758 \uc774\ud0c8 \uac00\ub2a5\uc131\uc774 \ub192\uc740 \uac83\uc73c\ub85c \ub098\ud0c0\ub0ac\uc2b5\ub2c8\ub2e4.<\/p>\n
\"query<\/p>\n
\uc544\ub798 \ucc28\ud2b8\ub294 \ud604\uc7ac \uc6d4 \ub9e4\ucd9c\uc774 $315\ub9cc\uc784\uc744 \ubcf4\uc5ec\uc90d\ub2c8\ub2e4. \uc774 \uc218\uc775\uc740 \uae30\uc874 \uace0\uac1d\uc758 \uc6d4 \uad6c\ub3c5 \ube44\uc6a9\uc774 $1\ub9cc\ud07c \uc99d\uac00\ud558\uba74 \uc57d $5\ub9cc \uc6d0, \uc6d4 \ube44\uc6a9\uc774 $2\ub9cc\ud07c \uc99d\uac00\ud558\uba74 \uc57d $23\ub9cc \uc6d0\uc774 \uc99d\uac00\ud560 \uac83\uc73c\ub85c \uc608\uce21\ub429\ub2c8\ub2e4. \uadf8\ub7ec\ub098 \uc6d4 \ube44\uc6a9\uc774 $3 \uc774\uc0c1 \uc99d\uac00\ud558\uba74 \uc608\uc0c1 \uc774\ud0c8\ub960\uc774 \ub192\uc544\uc838 \uc218\uc775\uc774 \uac10\uc18c\ud560 \uac83\uc73c\ub85c \uc608\uce21\ub429\ub2c8\ub2e4.<\/p>\n
\"querychart2\"
\n\uc774 \ubd84\uc11d\uc744 \ud1b5\ud574 \uc628\ub77c\uc778 \uc2a4\ud2b8\ub9ac\ubc0d \uc11c\ube44\uc2a4 \ud68c\uc0ac\uc758 \uc601\uc5c5\ud300\uc740 \uc774\ud0c8\ub960\uc744 \uc5b5\uc81c\ud558\uba74\uc11c \uc218\uc775\uc744 \uadf9\ub300\ud654\ud558\uae30 \uc704\ud574 \uc6d4 \uad6c\ub3c5\ub8cc\ub97c $2\ub85c \uc778\uc0c1\ud560 \uc218 \uc788\ub2e4\ub294 \uacb0\ub860\uc744 \ub0b4\ub9b4 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
\ucffc\ub9ac UI\uc758 '\ucc28\ud2b8 \ub2e4\uc6b4\ub85c\ub4dc' \uc635\uc158\uc744 \uc0ac\uc6a9\ud558\uc5ec \ucc28\ud2b8\ub97c \uc800\uc7a5\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
\uce74\uc6b0\uce58\ubca0\uc774\uc2a4 \uc560\ub110\ub9ac\ud2f1\uc2a4 \uc11c\ube44\uc2a4<\/h4>\n
\ud30c\uc774\uc36c SDK\uc758 Couchbase \ucffc\ub9ac API\ub97c \uc0ac\uc6a9\ud558\uc5ec Couchbase\uc5d0\uc11c \ub370\uc774\ud130\ub97c \uc77d\uc5c8\uc2b5\ub2c8\ub2e4. \ub300\uc2e0 Couchbase Analytics API\ub97c \uc0ac\uc6a9\ud558\ub824\ub294 \uacbd\uc6b0 \ub2e4\uc74c\uc740 Couchbase\uc5d0\uc11c \ub370\uc774\ud130\ub97c \uc77d\uace0 \ud310\ub2e4\uc2a4 \ub370\uc774\ud130 \ud504\ub808\uc784\uc5d0 \uc800\uc7a5\ud558\ub294 \uc608\uc81c\uc785\ub2c8\ub2e4.<\/p>\n
analytics_result = cb.analytics_query(\"SELECT customer.* FROM online_streaming customer\")\r\nanalytics_raw_data = pd.DataFrame(analytics_result)<\/pre>\n
\uce74\uc6b0\uce58\ubca0\uc774\uc2a4 \uc560\ub110\ub9ac\ud2f1\uc2a4 \uc11c\ube44\uc2a4\ub294 EDA, \ub370\uc774\ud130 \uc2dc\uac01\ud654 \ubc0f \ud559\uc2b5\ub41c ML \ubaa8\ub378(\uac1c\ubc1c\uc790 \ubbf8\ub9ac \ubcf4\uae30)\uc744 \uc2e4\ud589\ud558\ub294 \ub370\uc5d0\ub3c4 \uc0ac\uc6a9\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uc790\uc138\ud55c \ub0b4\uc6a9\uc740 <\/span>\uc560\ub110\ub9ac\ud2f1\uc2a4\uc6a9 N1QL \uc5b8\uc5b4 \ucc38\uc870<\/span><\/a> \uc5d0 \ub300\ud55c \uae30\uc0ac <\/span>\uce74\uc6b0\uce58\ubca0\uc774\uc2a4 \uc560\ub110\ub9ac\ud2f1\uc2a4 \ud30c\uc774\uc36c UDF\ub97c \uc0ac\uc6a9\ud558\uc5ec ML \ubaa8\ub378 \uc2e4\ud589\ud558\uae30<\/span><\/a> \uc5d0\uc11c \uc790\uc138\ud55c \ub0b4\uc6a9\uc744 \ud655\uc778\ud558\uc138\uc694.<\/span><\/p>\n
\uacb0\ub860<\/h4>\n
\uc774\ubc88 \uae00\uacfc \uc774\uc804 \uae00\uc5d0\uc11c Couchbase\uac00 \ub370\uc774\ud130 \uacfc\ud559\uc744 \uc27d\uac8c \ub9cc\ub4dc\ub294 \ubc29\ubc95\uc5d0 \ub300\ud574 \uc54c\uc544\ubcf4\uc558\uc2b5\ub2c8\ub2e4. \uace0\uac1d \uc774\ud0c8 \uc608\uce21\uc744 \uc608\ub85c \ub4e4\uc5b4, \ucffc\ub9ac \uc11c\ube44\uc2a4\ub97c \uc0ac\uc6a9\ud558\uc5ec \ud0d0\uc0c9\uc801 \ubd84\uc11d\uc744 \uc218\ud589\ud558\ub294 \ubc29\ubc95\uacfc Python SDK\ub97c \uc0ac\uc6a9\ud558\uc5ec \ub300\uaddc\ubaa8 \ud559\uc2b5 \ub370\uc774\ud130 \uc138\ud2b8\ub97c \ud6a8\uc728\uc801\uc73c\ub85c \uc77d\uace0 ML\uc5d0 \uc801\ud569\ud55c \ub370\uc774\ud130 \uad6c\uc870\ub85c \uc27d\uac8c \uc800\uc7a5\ud558\ub294 \ubc29\ubc95\uc744 \uc0b4\ud3b4\ubcf4\uc558\uc2b5\ub2c8\ub2e4.<\/p>\n
\ub610\ud55c ML \ubaa8\ub378, \uba54\ud0c0\ub370\uc774\ud130 \ubc0f \uc608\uce21\uc744 Couchbase\uc5d0 \uc800\uc7a5\ud558\ub294 \ubc29\ubc95\uacfc \uc608\uce21\uc744 \ubd84\uc11d\ud558\uae30 \uc704\ud574 \ucffc\ub9ac \ucc28\ud2b8\ub97c \uc0ac\uc6a9\ud558\ub294 \ubc29\ubc95\ub3c4 \uc0b4\ud3b4\ubd24\uc2b5\ub2c8\ub2e4.<\/p>\n
\uce74\uc6b0\uce58\ubca0\uc774\uc2a4 \ub370\uc774\ud130 \ud50c\ub7ab\ud3fc\uc740 \ucffc\ub9ac \ubc0f \ubd84\uc11d \uc11c\ube44\uc2a4\ub97c \uc2e4\ud589\ud558\ub294 \ud074\ub7ec\uc2a4\ud130\uc640 \ub3d9\uc77c\ud55c \ud074\ub7ec\uc2a4\ud130\uc5d0 \uc6d0\uc2dc \ub370\uc774\ud130, \uae30\ub2a5, ML \ubaa8\ub378, \ud574\ub2f9 \uba54\ud0c0\ub370\uc774\ud130 \ubc0f \uc608\uce21\uc744 \uc800\uc7a5\ud558\ub294 \ub370 \uc0ac\uc6a9\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uc774\ub807\uac8c \ud558\uba74 \ub370\uc774\ud130 \uacfc\ud559\uc5d0 \ud544\uc694\ud55c \ub3c4\uad6c\uc758 \uc218\ub97c \uc904\uc5ec \ud504\ub85c\uc138\uc2a4\ub97c \ube60\ub974\uace0 \uc27d\uac8c \uc9c4\ud589\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.<\/p>\n
\ub2e4\uc74c \ub2e8\uacc4<\/span><\/h4>\n
\uba38\uc2e0 \ub7ec\ub2dd\uacfc Couchbase\uc5d0 \ub300\ud574 \uc790\uc138\ud788 \uc54c\uc544\ubcf4\uace0 \uc2f6\ub2e4\uba74 \ub2e4\uc74c \ub2e8\uacc4\uc640 \ub9ac\uc18c\uc2a4\ub97c \ud1b5\ud574 \uc2dc\uc791\ud558\uc138\uc694:<\/span><\/p>\n