摩拜共享单车数据分析项目数据、代码、图表
资源内容介绍
摩拜共享单车数据分析项目的数据、代码、图表，基于上海摩拜单车的2016年8月份随机抽样大约10万条的开放订单数据进行分析 <html xmlns="http://www.w3.org/1999/xhtml"><meta charset="utf-8"><meta name="generator" content="pdf2htmlEX"><meta http-equiv="X-UA-Compatible" content="IE=edge,chrome=1"><link rel="stylesheet" href="/image.php?url=https://csdnimg.cn/release/download_crawler_static/css/base.min.css"><link rel="stylesheet" href="/image.php?url=https://csdnimg.cn/release/download_crawler_static/css/fancy.min.css"><link rel="stylesheet" href="/image.php?url=https://csdnimg.cn/release/download_crawler_static/12899807/raw.css"><script src="/image.php?url=https://csdnimg.cn/release/download_crawler_static/js/compatibility.min.js"></script><script src="/image.php?url=https://csdnimg.cn/release/download_crawler_static/js/pdf2htmlEX.min.js"></script><script>try{pdf2htmlEX.defaultViewer = new pdf2htmlEX.Viewer({});}catch(e){}</script><div id="sidebar" style="display: none"><div id="outline"></div></div><div id="pf1" class="pf w0 h0" data-page-no="1"><div class="pc pc1 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="/image.php?url=https://csdnimg.cn/release/download_crawler_static/12899807/bg1.jpg"><div class="c x1 y1 w2 h2"><div class="t m0 x2 h3 y2 ff1 fs0 fc0 sc0 ls0 ws0">摩拜共享单车数据分析项目报告</div><div class="t m0 x3 h3 y3 ff1 fs0 fc0 sc0 ls0 ws0">项目背景</div><div class="t m0 x3 h3 y4 ff1 fs0 fc0 sc0 ls0 ws0">数据探索</div><div class="t m0 x3 h3 y5 ff1 fs0 fc0 sc0 ls0 ws0">数据挖掘</div><div class="t m0 x3 h3 y6 ff1 fs0 fc0 sc0 ls0 ws0">数据分析</div><div class="t m0 x4 h3 y7 ff1 fs0 fc0 sc0 ls0 ws0">时间维度</div><div class="t m0 x4 h3 y8 ff1 fs0 fc0 sc0 ls0 ws0">空间维度</div><div class="t m0 x4 h3 y9 ff1 fs0 fc0 sc0 ls0 ws0">用户维度</div><div class="t m0 x5 h4 ya ff2 fs1 fc1 sc0 ls0 ws0"> </div><div class="t m0 x6 h5 yb ff3 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">摩</span><span class="fc8 sc0">拜</span></div><div class="t m0 x7 h5 yb ff4 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">共</span><span class="fc8 sc0">享</span></div><div class="t m0 x8 h5 yb ff5 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">单</span></div><div class="t m0 x9 h5 yb ff6 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">车</span></div><div class="t m0 xa h5 yb ff3 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">数</span><span class="fc8 sc0">据</span></div><div class="t m0 xb h5 yb ff4 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">分</span></div><div class="t m0 xc h5 yb ff3 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">析</span></div><div class="t m0 xd h5 yb ff7 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">项</span></div><div class="t m0 xe h5 yb ff8 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">目</span></div><div class="t m0 xf h5 yb ff3 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">报</span></div><div class="t m0 x10 h5 yb ff5 fs2 fc1 sc0 ls0 ws0"><span class="fc8 sc0">告</span></div><div class="t m0 x11 h6 yb ff9 fs2 fc0 sc0 ls0 ws0"> </div><div class="t m0 x5 h7 yc ffa fs3 fc1 sc0 ls0 ws0">项目背景<span class="_ _0"> </span><span class="ffb fc0"> </span></div><div class="t m0 x5 h8 yd ff1 fs1 fc1 sc0 ls0 ws0">随<span class="_ _1"></span>着<span class="_ _1"></span>智<span class="_ _1"></span>能<span class="_ _1"></span>手<span class="_ _1"></span>机<span class="_ _1"></span>的<span class="_ _1"></span>普<span class="_ _1"></span>及<span class="_ _1"></span>和<span class="_ _1"></span>手<span class="_ _1"></span>机<span class="_ _1"></span>用<span class="_ _1"></span>户<span class="_ _1"></span>的<span class="_ _1"></span>激<span class="_ _1"></span>增<span class="_ _1"></span>，<span class="_ _1"></span>共<span class="_ _1"></span>享<span class="_ _1"></span>单<span class="_ _1"></span>车<span class="_ _1"></span>作<span class="_ _1"></span>为<span class="_ _1"></span>城<span class="_ _1"></span>市<span class="_ _1"></span>交<span class="_ _1"></span>通<span class="_ _1"></span>系<span class="_ _1"></span>统<span class="_ _1"></span>的<span class="_ _1"></span>一<span class="_ _1"></span>个<span class="_ _1"></span>重<span class="_ _1"></span>要<span class="_ _1"></span>组<span class="_ _1"></span>成<span class="_ _1"></span>部<span class="_ _1"></span>分<span class="_ _1"></span>，<span class="_ _1"></span>以<span class="_ _1"></span>绿<span class="_ _1"></span>色<span class="_ _1"></span>环</div><div class="t m0 x5 h8 ye ff1 fs1 fc1 sc0 ls0 ws0">保、便捷<span class="_ _1"></span>高效、经济<span class="_ _1"></span>环保为特<span class="_ _1"></span>征蓬勃发<span class="_ _1"></span>展。共享单<span class="_ _1"></span>车企业通<span class="_ _1"></span>过在校园<span class="_ _1"></span>、公交站点<span class="_ _1"></span>、居民区<span class="_ _1"></span>、公共服<span class="_ _1"></span>务区</div><div class="t m0 x5 h8 yf ff1 fs1 fc1 sc0 ls0 ws0">等提供<span class="_"> </span>服<span class="_ _1"></span>务，完<span class="_ _1"></span>成交通行<span class="_ _1"></span>业最后<span class="_ _1"></span>一块<span class="ff2">“</span>拼<span class="_ _1"></span>图<span class="ff2">”</span>，<span class="_"> </span>与<span class="_ _1"></span>其他公<span class="_ _1"></span>共交通方<span class="_ _1"></span>式产生<span class="_ _1"></span>协同效应<span class="_ _1"></span>。共享<span class="_ _1"></span>单车有助<span class="_ _1"></span>于缓解</div><div class="t m0 x5 h8 y10 ff1 fs1 fc1 sc0 ls0 ws0">城市短<span class="_"> </span>距<span class="_ _1"></span>离交通<span class="_ _1"></span>出行和<span class="ff2">“<span class="_ _1"></span></span>最后一<span class="_ _1"></span>公里<span class="ff2">”</span>难<span class="_ _1"></span>题，但<span class="_"> </span>共<span class="_ _1"></span>享单车<span class="_ _1"></span>由于其运<span class="_ _1"></span>营特点<span class="_ _1"></span>，对企业<span class="_ _1"></span>在城市<span class="_ _1"></span>投放和调<span class="_ _1"></span>度单车</div><div class="t m0 x5 h8 y11 ff1 fs1 fc1 sc0 ls0 ws0">的规划管理方面，存在较大挑战。</div><div class="t m0 x5 h8 y12 ff1 fs1 fc1 sc0 ls0 ws0">基于<span class="_ _1"></span>上述<span class="_ _1"></span>背<span class="_ _1"></span>景，<span class="_ _1"></span>本文<span class="_ _1"></span>基<span class="_ _1"></span>于上<span class="_ _1"></span>海摩<span class="_ _1"></span>拜单<span class="_ _1"></span>车<span class="_ _1"></span>的<span class="ff2">2016<span class="_ _1"></span></span>年<span class="ff2">8<span class="_ _1"></span></span>月<span class="_ _1"></span>份随<span class="_ _1"></span>机抽<span class="_ _1"></span>样<span class="_ _1"></span>大约<span class="_ _1"></span><span class="ff2">10</span>万<span class="_ _1"></span>条的<span class="_ _1"></span>开<span class="_ _1"></span>放订<span class="_ _1"></span>单数<span class="_ _1"></span>据<span class="_ _1"></span>进行<span class="_ _1"></span>分析<span class="_ _1"></span>，</div><div class="t m0 x5 h8 y13 ff1 fs1 fc1 sc0 ls0 ws0">挖掘出数<span class="_ _1"></span>据背后的规<span class="_ _1"></span>律，用数<span class="_ _1"></span>据勾勒出<span class="_ _1"></span>摩拜共享单<span class="_ _1"></span>车的使用<span class="_ _1"></span>与用户出<span class="_ _1"></span>行现状，从<span class="_ _1"></span>而有助于<span class="_ _1"></span>摩拜共享<span class="_ _1"></span>单车</div><div class="t m0 x5 h8 y14 ff1 fs1 fc1 sc0 ls0 ws0">企业更好地推出营销策略，定位新单车的投放区域，调控车辆布置，更好地服务用户。</div><div class="t m0 x5 h7 y15 ffa fs3 fc1 sc0 ls0 ws0">数据探索<span class="_ _0"> </span><span class="ffb fc0"> </span></div><div class="t m0 x5 h8 y16 ff1 fs1 fc1 sc0 ls0 ws0">这一步中我们统观数据的全貌，对数据有个大体的了解，对数据进行质量探索和特征分析。</div><div class="t m0 x5 h8 y17 ff1 fs1 fc1 sc0 ls0 ws0">读取数据并查看数据集数据</div><div class="t m0 x5 h8 y18 ff1 fs1 fc1 sc0 ls0 ws0">该数<span class="_ _1"></span>据集<span class="_ _1"></span>为<span class="_ _1"></span>摩拜<span class="_ _1"></span>共享<span class="_ _1"></span>单<span class="_ _1"></span>车企<span class="_ _1"></span>业提<span class="_ _1"></span>供的<span class="_ _1"></span>上<span class="_ _1"></span>海城<span class="_ _1"></span>区<span class="ff2">2016<span class="_ _1"></span></span>年<span class="_ _1"></span><span class="ff2">8</span>月<span class="_ _1"></span>随机<span class="_ _1"></span>抽<span class="_ _1"></span>样的<span class="_ _1"></span><span class="ff2">10</span>万<span class="_ _1"></span>多条<span class="_ _1"></span>用<span class="_ _1"></span>户骑<span class="_ _1"></span>行用<span class="_ _1"></span>车<span class="_ _1"></span>数据<span class="_ _1"></span>（订<span class="_ _1"></span>单</div><div class="t m0 x5 h8 y19 ff1 fs1 fc1 sc0 ls0 ws0">数据），包含交易<span class="_ _1"></span>编号、用户<span class="ff2">ID</span>、车辆<span class="_ _1"></span><span class="ff2">ID</span>、骑行起点经纬<span class="_"> </span>度<span class="_ _1"></span>、骑行终点经纬度<span class="_ _1"></span>、租赁时间、还车时<span class="_ _1"></span>间、</div><div class="t m0 x5 h8 y1a ff1 fs1 fc1 sc0 ls0 ws0">骑行轨迹经纬度等数据。</div></div><div class="c x12 y1b w3 h9"><div class="t m0 x13 ha y1c ffc fs0 fc2 sc0 ls0 ws0">import<span class="fc1"> <span class="fc3">pandas</span> </span>as<span class="fc1"> <span class="fc3">pd</span></span></div><div class="t m0 x13 ha y1d ffc fs0 fc2 sc0 ls0 ws0">from<span class="fc1"> <span class="fc3">math</span> </span>import<span class="fc1"> <span class="fc3">radians</span>, <span class="fc3">cos</span>, <span class="fc3">sin</span>, <span class="fc3">asin</span>, <span class="fc3">sqrt</span>,<span class="fc3">ceil</span></span></div><div class="t m0 x13 ha y1e ffc fs0 fc2 sc0 ls0 ws0">import<span class="fc1"> <span class="fc3">numpy</span> </span>as<span class="fc1"> <span class="fc3">np</span></span></div><div class="t m0 x13 ha y1f ffc fs0 fc2 sc0 ls0 ws0">import<span class="fc1"> <span class="fc3">geohash</span></span></div><div class="t m0 x13 hb y20 ffc fs0 fc4 sc0 ls0 ws0">#<span class="ffd">数据读取</span></div><div class="t m0 x13 ha y21 ffc fs0 fc3 sc0 ls0 ws0">data<span class="fc1"> = </span>pd<span class="fc1">.</span>read_csv<span class="fc1">(<span class="fc5">"./mobike_shanghai_sample_updated.csv"</span>)</span></div><div class="t m0 x13 ha y22 ffc fs0 fc6 sc0 ls0 ws0">print<span class="fc1">(<span class="fc3">data</span>.<span class="fc3">head</span>(<span class="fc7">10</span>))</span></div></div><a class="l"><div class="d m1"></div></a><a class="l"><div class="d m1"></div></a><a class="l"><div class="d m1"></div></a><a class="l"><div class="d m1"></div></a><a class="l"><div class="d m1"></div></a><a class="l"><div class="d m1"></div></a><a class="l"><div class="d m1"></div></a><a class="l"><div class="d m1"></div></a><a class="l"><div class="d m1"></div></a><a class="l"><div class="d m1"></div></a><a class="l"><div class="d m1"></div></a></div><div class="pi" data-data='{"ctm":[1.611792,0.000000,0.000000,1.611792,0.000000,0.000000]}'></div></div></html><div id="pf2" class="pf w0 h0" data-page-no="2"><div class="pc pc2 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="/image.php?url=https://csdnimg.cn/release/download_crawler_static/12899807/bg2.jpg"><div class="c x1 y1 w2 h2"><div class="t m0 x5 h4 y23 ff2 fs1 fc1 sc0 ls0 ws0"> </div><div class="t m0 x5 h8 y24 ff1 fs1 fc1 sc0 ls0 ws0">查看<span class="_ _1"></span>数<span class="_ _1"></span>据集<span class="_ _1"></span>属<span class="_ _1"></span>性类<span class="_ _1"></span>型<span class="_ _1"></span>，在<span class="_ _1"></span>这<span class="_ _1"></span>里我<span class="_ _1"></span>们<span class="_ _1"></span>可以<span class="_ _1"></span>看<span class="_ _1"></span>出租<span class="_ _1"></span>赁<span class="_ _1"></span>时间<span class="_ _1"></span>和<span class="_ _1"></span>还车<span class="_ _1"></span>时<span class="_ _1"></span>间的<span class="_ _1"></span>数<span class="_"> </span>据<span class="_ _1"></span>类<span class="_ _1"></span>型为<span class="_ _1"></span><span class="ff2">object<span class="_ _1"></span></span>类型<span class="_ _1"></span>，<span class="_ _1"></span>我们<span class="_ _1"></span>紧<span class="_ _1"></span>接着</div><div class="t m0 x5 h8 y25 ff1 fs1 fc1 sc0 ls0 ws0">可以把它转化为<span class="ff2">datetime</span>类型</div></div><div class="c x12 y26 w3 hc"><div class="t m0 x13 ha y27 ffc fs0 fc6 sc0 ls0 ws0">print<span class="fc1">(<span class="fc3">data</span>.<span class="fc3">info</span>())</span></div></div><div class="c x12 y28 w3 hd"><div class="t m0 x13 ha y29 ffc fs0 fc3 sc0 ls0 ws0">data<span class="fc1">[<span class="fc5">'start_time'</span>] = </span>pd<span class="fc1">.</span>to_datetime<span class="fc1">(</span>data<span class="fc1">[<span class="fc5">'start_time'</span>])</span></div><div class="t m0 x13 ha y2a ffc fs0 fc3 sc0 ls0 ws0">data<span class="fc1">[<span class="fc5">'end_time'</span>] = </span>pd<span class="fc1">.</span>to_datetime<span class="fc1">(</span>data<span class="fc1">[<span class="fc5">'end_time'</span>])</span></div><div class="t m0 x13 ha y2b ffc fs0 fc6 sc0 ls0 ws0">print<span class="fc1">(<span class="fc3">data</span>.<span class="fc3">info</span>())</span></div></div></div><div class="pi" data-data='{"ctm":[1.611792,0.000000,0.000000,1.611792,0.000000,0.000000]}'></div></div><div id="pf3" class="pf w0 h0" data-page-no="3"><div class="pc pc3 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="/image.php?url=https://csdnimg.cn/release/download_crawler_static/12899807/bg3.jpg"><div class="c x1 y2c w2 h2"><div class="t m0 x5 h8 y2d ff1 fs1 fc1 sc0 ls0 ws0">查看数据集的空值分布情况，数据不存在空值</div><div class="t m0 x5 h4 y2e ff2 fs1 fc1 sc0 ls0 ws0"> </div><div class="t m0 x5 h7 y2f ffa fs3 fc1 sc0 ls0 ws0">数据挖掘<span class="_ _0"> </span><span class="ffb fc0"> </span></div><div class="t m0 x5 h8 y30 ff1 fs1 fc1 sc0 ls0 ws0">该数据集<span class="_ _1"></span>虽然多达<span class="ff2">10<span class="_ _1"></span></span>万多条用<span class="_ _1"></span>车交易数<span class="_ _1"></span>据，但每一<span class="_ _1"></span>条用车交<span class="_ _1"></span>易数据里<span class="_ _1"></span>只有<span class="ff2">10</span>个特<span class="_ _1"></span>征，并且<span class="_ _1"></span>该数据集<span class="_ _1"></span>和之</div><div class="t m0 x5 h8 y31 ff1 fs1 fc1 sc0 ls0 ws0">前<span class="_ _1"></span>的<span class="_ _1"></span>电<span class="_ _1"></span>影<span class="_ _1"></span>数<span class="_ _1"></span>据<span class="_ _1"></span>集<span class="_ _1"></span>不<span class="_ _1"></span>同<span class="_ _1"></span>，<span class="_ _1"></span>之<span class="_ _1"></span>前<span class="_ _1"></span>的<span class="_ _1"></span>电<span class="_ _1"></span>影<span class="_ _1"></span>数<span class="_ _1"></span>据<span class="_ _1"></span>集<span class="_ _1"></span>每<span class="_ _1"></span>一<span class="_ _1"></span>个<span class="_ _1"></span>特<span class="_ _1"></span>征<span class="_ _1"></span>都<span class="_ _1"></span>是<span class="_ _1"></span>独<span class="_ _1"></span>立<span class="_ _1"></span>的<span class="_ _1"></span>信<span class="_ _1"></span>息<span class="_ _1"></span>字<span class="_ _1"></span>段<span class="_ _1"></span>，<span class="_ _1"></span>不<span class="_ _1"></span>和<span class="_ _1"></span>其<span class="_ _1"></span>他<span class="_ _1"></span>特<span class="_ _1"></span>征<span class="_ _1"></span>产<span class="_ _1"></span>生<span class="_ _1"></span>明<span class="_ _1"></span>显<span class="_ _1"></span>关</div><div class="t m0 x5 h8 y32 ff1 fs1 fc1 sc0 ls0 ws0">联，而该<span class="_ _1"></span>数据集的特<span class="_ _1"></span>征之间拥<span class="_ _1"></span>有明显的<span class="_ _1"></span>相关性，我<span class="_ _1"></span>们可以通<span class="_ _1"></span>过关联组<span class="_ _1"></span>合特征之间<span class="_ _1"></span>的关系得<span class="_ _1"></span>到新的特<span class="_ _1"></span>征，</div><div class="t m0 x5 h8 y33 ff1 fs1 fc1 sc0 ls0 ws0">或者从一<span class="_ _1"></span>个特征反映<span class="_ _1"></span>出来的多<span class="_ _1"></span>个方面组<span class="_ _1"></span>合出多个新<span class="_ _1"></span>的特征扩<span class="_ _1"></span>充数据集<span class="_ _1"></span>，掌握事物<span class="_ _1"></span>的多个方<span class="_ _1"></span>面，挖掘<span class="_ _1"></span>出数</div><div class="t m0 x5 h8 y34 ff1 fs1 fc1 sc0 ls0 ws0">据更多潜在的规律，从而使得后面的数据分析可以进行更多维度的分析，得到对数据的更多认识。</div><div class="t m0 x5 h8 y35 ff1 fs1 fc1 sc0 ls0 ws0">如何发现新特征：</div><div class="t m0 x14 h8 y36 ff2 fs1 fc1 sc0 ls0 ws0">1<span class="_ _2"></span>. <span class="ff1">租赁时间</span> + <span class="ff1">还车时间</span> => <span class="ff1">骑行时长</span></div><div class="t m0 x14 h8 y37 ff2 fs1 fc1 sc0 ls0 ws0">2<span class="_ _2"></span>. <span class="ff1">骑行起点经纬度</span> + <span class="ff1">骑行终点经纬度</span> => <span class="ff1">骑行的位移</span></div><div class="t m0 x14 h8 y38 ff2 fs1 fc1 sc0 ls0 ws0">3<span class="_ _2"></span>. <span class="ff1">摩拜单车骑行轨迹经纬度</span> => <span class="ff1">骑行的路径</span></div><div class="t m0 x14 h8 y39 ff2 fs1 fc1 sc0 ls0 ws0">4<span class="_ _2"></span>. <span class="ff1">租赁<span class="_ _1"></span>时间</span> <span class="_ _1"></span>=> <span class="ff1">星期几<span class="_ _1"></span>（即每<span class="_ _1"></span>笔骑行订<span class="_ _1"></span>单发生<span class="_"> </span>在<span class="_ _1"></span>星期几<span class="_ _1"></span>）</span> + <span class="ff1">时<span class="_ _1"></span>间段（<span class="_ _1"></span>即每笔订<span class="_ _1"></span>单发生<span class="_ _1"></span>在一天</span>24<span class="_ _1"></span><span class="ff1">小时的<span class="_"> </span>哪</span></div><div class="t m0 x15 h8 y3a ff1 fs1 fc1 sc0 ls0 ws0">个时间段）</div><div class="t m0 x14 h8 y3b ff2 fs1 fc1 sc0 ls0 ws0">5<span class="_ _2"></span>. <span class="ff1">骑行时长</span> => <span class="ff1">订单金额（粗略估计</span>)</div><div class="t m0 x14 h8 y3c ff2 fs1 fc1 sc0 ls0 ws0">6<span class="_ _2"></span>. <span class="ff1">每笔订单金额</span> + <span class="ff1">每笔订单租赁时间</span> + <span class="ff1">每笔订单用户</span>ID=> <span class="ff1">用户分级（</span>RFM<span class="ff1">模型）</span></div><div class="t m0 x14 h8 y3d ff2 fs1 fc1 sc0 ls0 ws0">7<span class="_ _2"></span>. <span class="ff1">骑行起点终点经纬度</span> => <span class="ff1">骑行起点终点所处的地区</span></div><div class="t m0 x5 he y3e ffa fs1 fc1 sc0 ls0 ws0">租赁时间<span class="ffb"> + </span>还车时间<span class="ffb"> => </span>骑行时长</div><div class="t m0 x5 h8 y3f ff1 fs1 fc1 sc0 ls0 ws0">新增<span class="ff2">“lag”</span>列，通过开始时间<span class="ff2"> - </span>结束时间计算得到骑行时长，并把时长单位统一为分钟。</div></div><div class="c x12 y40 w3 hf"><div class="t m0 x13 ha y41 ffc fs0 fc6 sc0 ls0 ws0">print<span class="fc1">(<span class="fc3">data</span>.<span class="fc3">isnull</span>().<span class="fc3">sum</span>())</span></div></div><div class="c x12 y42 w3 hc"><div class="t m0 x13 ha y43 ffc fs0 fc3 sc0 ls0 ws0">data<span class="fc1">[<span class="fc5">"lag"</span>] = (</span>data<span class="fc1">.</span>end_time<span class="fc1"> <span class="fc9">-</span> </span>data<span class="fc1">.</span>start_time<span class="fc1">).</span>dt<span class="fc1">.</span>seconds<span class="fc9">/<span class="fc7">60</span></span></div></div><a class="l"><div class="d m1"></div></a></div><div class="pi" data-data='{"ctm":[1.611792,0.000000,0.000000,1.611792,0.000000,0.000000]}'></div></div><div id="pf4" class="pf w0 h0" data-page-no="4"><div class="pc pc4 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="/image.php?url=https://csdnimg.cn/release/download_crawler_static/12899807/bg4.jpg"><div class="c x1 y44 w2 h10"><div class="t m0 x5 he y45 ffa fs1 fc1 sc0 ls0 ws0">骑行起点经纬度<span class="ffb"> + </span>骑行终点经纬度<span class="ffb"> => </span>骑行的位移</div><div class="t m0 x5 h8 y46 ff1 fs1 fc1 sc0 ls0 ws0">新增<span class="ff2">“distance”</span>列，通过计算骑行起点和终点的经纬度得到骑行的位移<span class="ff2">,</span>单位为千米。</div><div class="t m0 x5 h8 y47 ff2 fs1 fc1 sc0 ls0 ws0">geodistance()<span class="ff1">封<span class="_ _1"></span>装<span class="_ _1"></span>的<span class="_"> </span>是<span class="_ _1"></span>通<span class="_ _1"></span>过<span class="_ _1"></span>两点<span class="_ _1"></span>经<span class="_ _1"></span>纬度<span class="_ _1"></span>求<span class="_ _1"></span>两<span class="_ _1"></span>点直<span class="_ _1"></span>线<span class="_ _1"></span>距离<span class="_ _1"></span>的<span class="_ _1"></span>数<span class="_ _1"></span>学公<span class="_ _1"></span>式<span class="_ _1"></span>。对<span class="_ _1"></span>背<span class="_ _1"></span>后<span class="_ _1"></span>公式<span class="_ _1"></span>的<span class="_ _1"></span>推导<span class="_ _1"></span>过<span class="_ _1"></span>程<span class="_ _1"></span>和数<span class="_ _1"></span>学<span class="_ _1"></span>原理</span></div><div class="t m0 x5 h8 y48 ff1 fs1 fc1 sc0 ls0 ws0">感兴趣的可以戳<span class="ff2"> <span class="fc0">https://blog.csdn.net/sunjianqiang12345/article/details/60393437</span> </span>了解。</div><div class="t m0 x5 he y49 ffa fs1 fc1 sc0 ls0 ws0">摩拜单车骑行轨迹经纬度<span class="ffb"> => </span>骑行的路径</div><div class="t m0 x5 h8 y4a ff1 fs1 fc1 sc0 ls0 ws0">新增<span class="ff2">“adderLength”</span>列，通过计算<span class="ff2">“track”</span>列数据得到骑行路径。</div><div class="t m0 x5 h8 y4b ff1 fs1 fc1 sc0 ls0 ws0">把骑行轨迹字符串按<span class="ff2">“#”</span>分隔符分隔成列表，得到骑行轨迹的经纬度信息列表</div><div class="t m0 x5 h8 y4c ff1 fs1 fc1 sc0 ls0 ws0">通过不断<span class="_ _1"></span>两两轮询该<span class="_ _1"></span>骑行轨迹<span class="_ _1"></span>列表，每<span class="_ _1"></span>一次按顺序<span class="_ _1"></span>取出两个<span class="_ _1"></span>列表元素<span class="_ _1"></span>（前轨迹采<span class="_ _1"></span>样点的经<span class="_ _1"></span>纬度和后<span class="_ _1"></span>轨迹</div><div class="t m0 x5 h8 y4d ff1 fs1 fc1 sc0 ls0 ws0">采样点的经纬度）<span class="_ _1"></span>按<span class="ff2">“,”</span>分隔符分隔，得<span class="_ _1"></span>到四个值（前轨迹<span class="_"> </span>采<span class="_ _1"></span>样点的经度、前轨<span class="_ _1"></span>迹采样点的纬度、后<span class="_ _1"></span>轨迹</div><div class="t m0 x5 h8 y4e ff1 fs1 fc1 sc0 ls0 ws0">采样点的经度、后轨迹采样点<span class="_"> </span>的<span class="_"> </span>纬<span class="_ _1"></span>度），组合成一个<span class="ff2">item</span>字典传入给<span class="ff2">geodistance()<span class="_ _1"></span></span>计算出两点的位移，再</div><div class="t m0 x5 h8 y4f ff1 fs1 fc1 sc0 ls0 ws0">把每一小段每一小段的位移累加起来得到骑行的路径。</div></div><div class="c x12 y50 w3 h11"><div class="t m0 x13 ha y51 ffc fs0 fc2 sc0 ls0 ws0">def<span class="fc1"> <span class="fca">geodistance</span>(<span class="fc3">item</span>):</span></div><div class="t m0 x13 ha y52 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">lng1_r</span>, <span class="fc3">lat1_r</span>, <span class="fc3">lng2_r</span>, <span class="fc3">lat2_r</span> = <span class="fc6">map</span>(<span class="fc3">radians</span>, [<span class="fc3">item</span>[<span class="fc5">"start_location_x"</span>], </div><div class="t m0 x13 hb y53 ffc fs0 fc3 sc0 ls0 ws0">item<span class="fc1">[<span class="fc5">"start_location_y"</span>], </span>item<span class="fc1">[<span class="fc5">"end_location_x"</span>], </span>item<span class="fc1">[<span class="fc5">"end_location_y<span class="ffd">，</span>"</span>]]) <span class="fc4"># <span class="ffd">经纬度转</span></span></span></div><div class="t m0 x13 hb y54 ffd fs0 fc4 sc0 ls0 ws0">换成弧度</div><div class="t m0 x13 ha y55 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">dlon</span> = <span class="fc3">lng1_r</span> <span class="fc9">-</span> <span class="fc3">lng2_r</span></div><div class="t m0 x13 ha y56 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">dlat</span> = <span class="fc3">lat1_r</span> <span class="fc9">-</span> <span class="fc3">lat2_r</span></div><div class="t m0 x13 ha y57 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">dis</span> = <span class="fc3">sin</span>(<span class="fc3">dlat<span class="fc9">/<span class="fc7">2</span></span></span>)<span class="fc9">**<span class="fc7">2</span></span> <span class="fc9">+</span> <span class="fc3">cos</span>(<span class="fc3">lat1_r</span>) <span class="fc9">*</span> <span class="fc3">cos</span>(<span class="fc3">lat2_r</span>) <span class="fc9">*</span> <span class="fc3">sin</span>(<span class="fc3">dlon<span class="fc9">/<span class="fc7">2</span></span></span>)<span class="fc9">**<span class="fc7">2</span></span></div><div class="t m0 x13 hb y58 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">distance</span> = <span class="fc7">2</span> <span class="fc9">*</span> <span class="fc3">asin</span>(<span class="fc3">sqrt</span>(<span class="fc3">dis</span>)) <span class="fc9">*</span> <span class="fc7">6371</span> <span class="fc9">*</span> <span class="fc7">1000</span> <span class="fc4"># <span class="ffd">地球平均半径为</span>6371km</span></div><div class="t m0 x13 ha y59 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">distance</span> = <span class="fc6">round</span>(<span class="fc3">distance<span class="fc9">/<span class="fc7">1000</span></span></span>,<span class="fc7">3</span>)</div><div class="t m0 x13 ha y5a ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc2">return</span> <span class="fc3">distance</span></div><div class="t m0 x13 hb y5b ffc fs0 fc4 sc0 ls0 ws0">#data<span class="ffd">按行应用</span>geodistance()<span class="ffd">得到</span>distance<span class="ffd">列的数值</span></div><div class="t m0 x13 ha y5c ffc fs0 fc3 sc0 ls0 ws0">data<span class="fc1">[<span class="fc5">"distance"</span>] = </span>data<span class="fc1">.</span>apply<span class="fc1">(</span>geodistance<span class="fc1">,</span>axis<span class="fc1">=<span class="fc7">1</span>)</span></div></div><div class="c x12 y5d w3 h12"><div class="t m0 x13 hb y5e ffc fs0 fc4 sc0 ls0 ws0">#<span class="ffd">通过摩拜单车的踪迹获取每次交易骑行的路径</span></div><div class="t m0 x13 ha y5f ffc fs0 fc2 sc0 ls0 ws0">def<span class="fc1"> <span class="fca">geoaadderLength</span>(<span class="fc3">item</span>):</span></div><div class="t m0 x13 ha y60 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">track_list</span> = <span class="fc3">item</span>[<span class="fc5">"track"</span>].<span class="fc3">split</span>(<span class="fc5">"#"</span>)</div><div class="t m0 x13 ha y61 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">adderLength_item</span> = {}</div><div class="t m0 x13 ha y62 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">adderLength</span> = <span class="fc7">0</span></div><div class="t m0 x13 ha y63 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc2">for</span> <span class="fc3">i</span> <span class="fc2">in</span> <span class="fc6">range</span>(<span class="fc6">len</span>(<span class="fc3">track_list</span>)<span class="fc9">-<span class="fc7">1</span></span>):</div><div class="t m0 x13 ha y64 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">start_loc</span> = <span class="fc3">track_list</span>[<span class="fc3">i</span>].<span class="fc3">split</span>(<span class="fc5">","</span>)</div><div class="t m0 x13 ha y65 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">end_loc</span> = <span class="fc3">track_list</span>[<span class="fc3">i<span class="fc9">+<span class="fc7">1</span></span></span>].<span class="fc3">split</span>(<span class="fc5">","</span>)</div><div class="t m0 x13 ha y66 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">adderLength_item</span>[<span class="fc5">"start_location_x"</span>],<span class="fc3">adderLength_item</span>[<span class="fc5">"start_location_y"</span>] = </div><div class="t m0 x13 ha y67 ffc fs0 fc6 sc0 ls0 ws0">float<span class="fc1">(<span class="fc3">start_loc</span>[<span class="fc7">0</span>]),</span>float<span class="fc1">(<span class="fc3">start_loc</span>[<span class="fc7">1</span>])</span></div><div class="t m0 x13 ha y68 ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">adderLength_item</span>[<span class="fc5">"end_location_x"</span>],<span class="fc3">adderLength_item</span>[<span class="fc5">"end_location_y"</span>] = </div><div class="t m0 x13 ha y69 ffc fs0 fc6 sc0 ls0 ws0">float<span class="fc1">(<span class="fc3">end_loc</span>[<span class="fc7">0</span>]),</span>float<span class="fc1">(<span class="fc3">end_loc</span>[<span class="fc7">1</span>])</span></div></div><a class="l"><div class="d m1"></div></a></div><div class="pi" data-data='{"ctm":[1.611792,0.000000,0.000000,1.611792,0.000000,0.000000]}'></div></div><div id="pf5" class="pf w0 h0" data-page-no="5"><div class="pc pc5 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="/image.php?url=https://csdnimg.cn/release/download_crawler_static/12899807/bg5.jpg"><div class="c x1 y44 w2 h10"><div class="t m0 x5 he y6a ffa fs1 fc1 sc0 ls0 ws0">租赁时间<span class="ffb"> => </span>星期几<span class="ffb"> + </span>时间段（<span class="ffb">24</span>小时制）</div><div class="t m0 x5 h8 y6b ff1 fs1 fc1 sc0 ls0 ws0">新<span class="_ _1"></span>增<span class="_"> </span><span class="ff2">“weekday”<span class="_ _1"></span></span>（<span class="_ _1"></span>即<span class="_ _1"></span>每<span class="_ _1"></span>笔<span class="_ _1"></span>骑<span class="_ _1"></span>行<span class="_ _1"></span>订<span class="_ _1"></span>单<span class="_ _1"></span>发<span class="_ _1"></span>生<span class="_ _1"></span>在<span class="_ _1"></span>星<span class="_ _1"></span>期<span class="_ _1"></span>几）<span class="_ _1"></span>和<span class="_ _1"></span><span class="ff2">“hour”<span class="_ _1"></span></span>（<span class="_ _1"></span>即<span class="_ _1"></span>每<span class="_ _1"></span>笔<span class="_ _1"></span>订<span class="_ _1"></span>单<span class="_ _1"></span>发<span class="_ _1"></span>生<span class="_ _1"></span>在<span class="_ _1"></span>一<span class="_ _1"></span>天<span class="_"> </span><span class="ff2">24<span class="_ _1"></span></span>小<span class="_ _1"></span>时<span class="_ _1"></span>的<span class="_ _1"></span>哪<span class="_ _1"></span>个<span class="_ _1"></span>时<span class="_ _1"></span>间</div><div class="t m0 x5 h8 y6c ff1 fs1 fc1 sc0 ls0 ws0">段）</div><div class="t m0 x5 he y6d ffa fs1 fc1 sc0 ls0 ws0">骑行时长<span class="ffb"> => </span>订单金额</div><div class="t m0 x5 h8 y6e ff1 fs1 fc1 sc0 ls0 ws0">新增<span class="_ _1"></span><span class="ff2">“cost”<span class="_ _1"></span></span>列，<span class="_ _1"></span>根据<span class="_ _1"></span>每<span class="_ _1"></span>笔订<span class="_ _1"></span>单的<span class="_ _1"></span>骑<span class="_ _1"></span>行时<span class="_ _1"></span>长<span class="_"> </span>，<span class="_ _1"></span>粗<span class="_ _1"></span>略估<span class="_ _1"></span>计<span class="_ _1"></span>订单<span class="_ _1"></span>金额<span class="_ _1"></span>，<span class="_ _1"></span>参照<span class="_ _1"></span><span class="ff2">2016<span class="_ _1"></span></span>年摩<span class="_ _1"></span>拜收<span class="_ _1"></span>费<span class="_ _1"></span>标准<span class="_ _1"></span>，按<span class="_ _1"></span>每<span class="_ _1"></span><span class="ff2">30</span>分<span class="_ _1"></span>钟</div><div class="t m0 x5 h8 y6f ff1 fs1 fc1 sc0 ls0 ws0">收取<span class="ff2">1</span>元。</div><div class="t m0 x5 he y70 ffa fs1 fc1 sc0 ls0 ws0">订单金额<span class="ffb"> => </span>用户分级（<span class="ffb">RFM</span>模型）</div><div class="t m0 x5 h8 y71 ff1 fs1 fc1 sc0 ls0 ws0">由于我<span class="_ _1"></span>们拥有了<span class="_ _1"></span>每笔交<span class="_ _1"></span>易的用户<span class="_ _1"></span><span class="ff2">id</span>、消<span class="_ _1"></span>费金额<span class="_ _1"></span>、消费时<span class="_ _1"></span>间，我<span class="_ _1"></span>们可以考<span class="_ _1"></span>虑运用<span class="_ _1"></span><span class="ff2">RFM</span>模型<span class="_ _1"></span>，对用户<span class="_ _1"></span>进行分</div><div class="t m0 x5 h8 y72 ff1 fs1 fc1 sc0 ls0 ws0">级，这里采取的模型是<span class="ff2">RFM</span>模型。</div><div class="t m0 x5 h8 y73 ff2 fs1 fc1 sc0 ls0 ws0">RFM<span class="ff1">模型是进行用户价值细分的一种方法，是用以研究用户的数学模型。</span></div><div class="t m0 x15 h8 y74 ff2 fs1 fc1 sc0 ls0 ws0">R<span class="ff1">（</span>Recency<span class="ff1">）最近一次消费时间：表示用户最近一次消费距离现在的时间；</span></div><div class="t m0 x15 h8 y75 ff2 fs1 fc1 sc0 ls0 ws0">F<span class="ff1">（</span>Frequency<span class="ff1">）消费频率：消费频率是指用户在统计周期内购买商品的次数；</span></div><div class="t m0 x15 h8 y76 ff2 fs1 fc1 sc0 ls0 ws0">M<span class="ff1">（</span>Monetary<span class="ff1">）<span class="_ _1"></span>消费金额<span class="_ _1"></span>：消费金<span class="_ _1"></span>额是指用<span class="_ _1"></span>户在统计<span class="_ _1"></span>周期内消<span class="_ _1"></span>费的总金<span class="_ _1"></span>额，体现<span class="_ _1"></span>了消费<span class="_"> </span>者<span class="_ _1"></span>为企业创<span class="_ _1"></span>利</span></div><div class="t m0 x15 h8 y77 ff1 fs1 fc1 sc0 ls0 ws0">的多少；</div><div class="t m0 x5 h8 y78 ff1 fs1 fc1 sc0 ls0 ws0">这<span class="ff2">3</span>个维度，帮助我们把用户划分为标准的<span class="ff2">8</span>类</div></div><div class="c x12 y79 w3 h13"><div class="t m0 x13 ha y7a ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">adderLength_each</span> = <span class="fc3">geodistance</span>(<span class="fc3">adderLength_item</span>)</div><div class="t m0 x13 ha y7b ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc3">adderLength</span> = <span class="fc3">adderLength_each</span> <span class="fc9">+</span> <span class="fc3">adderLength</span></div><div class="t m0 x13 ha y7c ffc fs0 fc1 sc0 ls0 ws0"> <span class="fc2">return</span> <span class="fc3">adderLength</span></div><div class="t m0 x13 ha y7d ffc fs0 fc3 sc0 ls0 ws0">data<span class="fc1">[<span class="fc5">"adderLength"</span>] = </span>data<span class="fc1">.</span>apply<span class="fc1">(</span>geoaadderLength<span class="fc1">,</span>axis<span class="fc1">=<span class="fc7">1</span>)</span></div></div><div class="c x12 y7e w3 h14"><div class="t m0 x13 ha y7f ffc fs0 fc3 sc0 ls0 ws0">data<span class="fc1">[<span class="fc5">'weekday'</span>] = </span>data<span class="fc1">.</span>start_time<span class="fc1">.</span>apply<span class="fc1">(<span class="fc2">lambda</span> </span>x<span class="fc1">: </span>x<span class="fc1">.</span>isoweekday<span class="fc1">())</span></div><div class="t m0 x13 ha y80 ffc fs0 fc3 sc0 ls0 ws0">data<span class="fc1">[<span class="fc5">'hour'</span>] = </span>data<span class="fc1">.</span>start_time<span class="fc1">.</span>apply<span class="fc1">(<span class="fc2">lambda</span> </span>x<span class="fc1">: </span>x<span class="fc1">.</span>utctimetuple<span class="fc1">().</span>tm_hour<span class="fc1">)</span></div></div><div class="c x12 y81 w3 h15"><div class="t m0 x13 ha y82 ffc fs0 fc3 sc0 ls0 ws0">data<span class="fc1">[<span class="fc5">'cost'</span>] = </span>data<span class="fc1">.</span>lag<span class="fc1">.</span>apply<span class="fc1">(<span class="fc2">lambda</span> </span>x<span class="fc1">: </span>ceil<span class="fc1">(</span>x<span class="fc9">/<span class="fc7">30<span class="fc1">))</span></span></span></div></div></div><div class="pi" data-data='{"ctm":[1.611792,0.000000,0.000000,1.611792,0.000000,0.000000]}'></div></div>