最详细的 Python 结合 RFM 模型实现用户分层实操案例！

{"type":"doc","content":[{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"作者：Cherich_sun","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"链接：","attrs":{}},{"type":"link","attrs":{"href":"https://www.jianshu.com/p/f020dfdce58d","title":"","type":null},"content":[{"type":"text","text":"https://www.jianshu.com/p/f020dfdce58d","attrs":{}}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"本文为读者投稿","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"公众号「","attrs":{}},{"type":"codeinline","content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"杰哥的IT之旅","attrs":{}}],"attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"」，后台回复：「","attrs":{}},{"type":"codeinline","content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"RFM数据","attrs":{}}],"attrs":{}},{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"」即可获取本文完整数据。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"原文链接：","attrs":{}},{"type":"link","attrs":{"href":"https://mp.weixin.qq.com/s/YOIhJERCfs19xIFRA44cpQ","title":"","type":null},"content":[{"type":"text","text":"最详细的 Python 结合 RFM 模型实现用户分层实操案例！","attrs":{}}]}]}],"attrs":{}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"写在最前：做数据分析的小伙伴可能多多少少都知道一些分析方法，但是谈到分析思维却没有底气或者遇到业务问题，不知道如何下手。如果你有上述困惑，那么本篇文章可以作为参考。下图是整理的分析方法论及方法。如果能够灵活运用，将能够解决工作中 80% 以上问题。注意的是，方法论是思维层面，方法是执行层面。那么，重点是我们如何将其应用到实际业务中。本文将以 RFM 模型 为例，运用到实际案例中。（本文以 Python 实现，Excel 也可以。）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/12/12344c02e396dbe4a255334b343a2d0f.png","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"项目背景：某生鲜外卖APP于2018年1月1日成立，主营新鲜蔬菜瓜果，海鲜肉禽。APP上线后，市场推广期为一年。通过分析发现原来几个重要的客户被竞争对手挖走了，而这几个用户对平台贡献了80%的销售额。之前对所有用户采用一样的运营策略，为了解决这个问题，需要对用户进行分类，了解当前用户分层情况，进行精细化运营。","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"一、整体分析流程","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"1、分析目的：用户分类","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"2、数据获取：Excel 数据","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"3、清洗加工：Excel、Python","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"4、建立模型：RFM","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"5、数据可视化","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"6、结论与建议","attrs":{}}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"二、RFM 模型的理解","attrs":{}}]},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/a3/a34bcbf730ac1b158995c3ddfdc9ea27.png","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/8a/8a4711bac7b3fafd03f3f4fdf0ec26e0.png","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"最终将 RFM 模型处理后的结果，作为用户标签，帮助运营更精准地制定活动规则以提升用户使用黏性，强化用户感知。最终实现的效果图如下：","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/f5/f507cd6b9d9f28fb4c436b1266cb78f6.png","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"三、利用 Python 实现 RFM 用户分层","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"1、获取数据","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"python"},"content":[{"type":"text","text":"import pandas as pd\ndata = pd.read_excel('C:/Users/cherich/Desktop/用户信息.xlsx')\ndata.head()","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/97/97f8b1560d0e643b6ed1376f8b760879.png","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"python"},"content":[{"type":"text","text":"data.info()","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/69/698a4f96b4572c7c354a5f6967641134.png","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"说明：当前数据集是5000条用户数据，存在缺失值对本次分析不会造成影响。数据清洗，通常包括处理缺失值、重复值、转换数据类型三种。所以仅考虑数据类型即可。这里有个前提条件，R、F、M 应该有一个参照时间，如果活动持续到现在，可以截止到现在。但是我们的数据是历史数据，所以需要查找活动结束时间。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"python"},"content":[{"type":"text","text":"data.sort_values(by='最后一次成交', ascending=False)","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/a0/a041eb0b3da5dc086178120ab4a5333b.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"2、数据处理","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"python"},"content":[{"type":"text","text":"# 活动结束时间 2019-06-30\ndata['最后一次成交']=data['最后一次成交'].astype('str')\nstop_date = pd.to_datetime('2019-06-30')\ndatas = data.drop(columns=['注册时间','会员开通时间','会员类型','城市','区域','最后一次登陆'])\ndatas['最后一次成交时间'] = datas['最后一次成交'].apply(lambda x:x.split()[0])\n\ndatas['最后一次成交时间'] = pd.to_datetime(datas['最后一次成交时间'])\n\ndatas['R1'] = datas['最后一次成交时间'].apply(lambda x:stop_date-x)\n\ndatas['F1'] = datas['非会员累计购买次数']+datas['会员累计购买次数']\n\ndatas['M1'] = datas['非会员累计消费'] + datas['会员累计消费']\ndatas['R1']= datas['R1'].astype(str)\ndatas['R1']= datas['R1'].apply(lambda x:x.split()[0])\ndatas","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/df/dfef85975b14c1903e282bf3db694e97.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"说明：以上操作目的是将R指标由时间类型转换成可计算格式，为接下来建立模型，计算时间间隔做准备。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"3、建立模型","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"建立模型，需要分别对F、R、M 分别计算各自的平均值。但是要注意三个指标数据存在极大值、极小值的情况，这对结果会产生一定的误差，所以解决方案是将其标准化，设置分段区间，5分制，5分为最高。（数值区间可根据具体业务灵活调整或者用四分位数）","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/55/556dba1c44a9549da031bbbee06bc504.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"python"},"content":[{"type":"text","text":"def R_score(n):\n n = int(n)\n if 0 R_mean else 0)\n\ndatas['F'] = datas['F1_score'].apply(lambda x: 1 if x> F_mean else 0)\n\ndatas['M'] = datas['M1_score'].apply(lambda x: 1 if x> M_mean else 0)\ndatas","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/14/14da39e2b82a262641479d33de04b2a7.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"python"},"content":[{"type":"text","text":"datas['RFM'] = datas['R'].apply(str)+datas['F'].apply(str)+datas['M'].apply(str)\ndatas\n\ndef user_tag(rfm):\n if rfm=='000':\n res = '流失用户'\n elif rfm=='010':\n res = '一般维持用户'\n elif rfm=='100':\n res = '新客户'\n elif rfm=='110':\n res = '潜力客户'\n elif rfm=='001':\n res = '重要挽留客户'\n elif rfm=='101':\n res = '重要深耕客户'\n elif rfm=='011':\n res = '重要唤回客户'\n else:\n res = '重要价值客户'\n return res\ndatas['user_tag']=datas['RFM'].apply(user_tag)\ndatas","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/c0/c0a130beb61eda5283edcdbe981df1b1.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"4、数据可视化","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"python"},"content":[{"type":"text","text":"import matplotlib.pyplot as plt\nimport seaborn as sns\nimport matplotlib as mpl\nsns.set(font='SimHei',style='darkgrid')\n\nuser_tag = datas.groupby(datas['user_tag']).size()\n\nplt.figure(figsize = (10,4),dpi=80)\n\nuser_tag.sort_values(ascending=True,inplace=True)\n\nplt.title(label='生鲜平台用户分层对比',\n fontsize=22, color='white',\n backgroundcolor='#334f65', pad=20)\n\ns = plt.barh(user_tag.index,user_tag.values , height=0.8, color=plt.cm.coolwarm_r(np.linspace(0,1,len(user_tag))))\nfor rect in s:\n width = rect.get_width()\n plt.text(width+40,rect.get_y() + rect.get_height()/2, str(width),ha= 'center')\n\nplt.grid(axis='y')\nplt.show()","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/3c/3c5338e3a216621bf2a3f718e1eff972.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"codeblock","attrs":{"lang":"python"},"content":[{"type":"text","text":"groups_b = datas.groupby(by='user_tag').size()\n\nplt.figure(figsize = (10,6),dpi=80)\nplt.title(label='生鲜平台用户分层占比',\n fontsize=22, color='white',\n backgroundcolor='#334f65', pad=20)\n\nexplodes = [0.6, 0, 0, 0, 0,0,0.4,0.8]\n\npatches, l_text, p_text = plt.pie(groups_b.values,labels = groups_b.index, shadow=True,colors=plt.cm.coolwarm_r(np.linspace(0,1,len(groups_b))), autopct='%.2f%%', explode=explodes,startangle=370)\nplt.legend(ins,bbox_to_anchor=(2, 1.0))\nplt.show()","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/91/91942d7b715369bc6075355126069596.png","alt":null,"title":null,"style":[{"key":"width","value":"75%"},{"key":"bordertype","value":"none"}],"href":null,"fromPaste":true,"pastePass":true}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"5、结论与建议","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"以上基本完成了RFM模型实现用户分层，可以看出新客户占比30%左右，重要价值客户占比30%左右。两者是平台的最主要用户类型。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"接下来就需要结合具体业务来制定运营策略。最后分享的是，现在我们看到最多的招聘需求是具备分析思维。那什么是分析思维。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我的理解是，首先要理解业务，其次要掌握分析方法，要明确分析方法存在的意义是帮助我们将零散业务问题归类，归类的过程形成分析思路，有了分析思路，那你就具备了分析思维。","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong","attrs":{}}],"text":"原创不易，码字不易。 觉得这篇文章对你有点用的话，麻烦你为本文点个赞，留言或转发一下，因为这将是我输出更多优质文章的动力，感谢！","attrs":{}}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"image","attrs":{"src":"https://static001.geekbang.org/infoq/01/019b65e5ec6661e8bbcaec704494797d.jpeg","alt":null,"title":"","style":[{"key":"width","value":"100%"},{"key":"bordertype","value":"none"}],"href":"","fromPaste":false,"pastePass":false}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}}]}