week3zhangruochen

week3大作业提交/zhangruochen/homework1.py

@@ -0,0 +1,92 @@
+
+# coding: utf-8
+
+# In[1]:
+
+#一定时间内北京各城区发贴量最多的三大类目
+
+
+# In[1]:
+
+import pymongo
+import charts
+from datetime import timedelta,date
+
+
+# In[2]:
+
+client = pymongo.MongoClient('localhost', 27017)
+test = client['test']
+homework = test['homework']
+homework_output = test['homework_output']
+
+In[18]:
+
+for i in homework.find():
+    if (i['area']==[]) or (i['area'] == None):
+        area_dist = None
+    else:
+        area_dist = i['area'][0]
+    homework.update_one({'_id':i['_id']},{'$set':{'area_dist':area_dist}})
+
+In[24]:
+
+for i in homework.find():
+    frags = i['pub_date'].split('.')
+    if len(frags) == 1:
+        date = frags[0]
+    else:
+        date = '%s-%s-%s'%(frags[0],frags[1],frags[2])
+    homework.update_one({'_id':i['_id']},{'$set':{'pub_date':date}})
+
+
+# In[3]:
+
+def get_all_dates(date1,date2):
+    the_date = date(int(date1.split('-')[0]),int(date1.split('-')[1]),int(date1.split('-')[2]))
+    end_date = date(int(date2.split('-')[0]),int(date2.split('-')[1]),int(date2.split('-')[2]))
+    days = timedelta(days=1)
+    while the_date <= end_date:
+        yield (str(the_date)) 
+        the_date += days
+
+# In[4]:
+
+def get_data_within(date1,date2,cates,area_dist):
+    data_list = []
+    for cate in cates:
+        total_post = 0
+        for date in get_all_dates(date1,date2):
+            a = list(homework.find({'area_dist':area_dist,'cates':cate,'pub_date':date}))
+            each_day_post = len(a)
+            total_post += each_day_post
+#         print(area_dist,cate,total_post)
+        data_ori = {
+        'name':cate,
+        'data':[total_post],
+        'type':'column'
+    }
+        data_list.append(data_ori)
+    data_list.sort(key=lambda x:x['data'][0])
+    data = data_list[-3:]
+    for i in data:
+        yield i
+#     print(data_list)
+
+
+# In[5]:
+
+cates=[]
+dists=[]
+for i in homework.find():
+    cates.append(i['cates'])
+    dists.append(i['area_dist'])
+cates_index = list(set(cates))
+dists_index = list(set(dists))
+# print(cates_index,dists_index)
+
+
+# In[6]:
+for dist in dists_index:
+    series = [i for i in get_data_within('2015-12-21','2016-01-15',cates_index,dist)]
+    charts.plot(series, options=dict(title=dict(text='各大区排名前三类目及发帖量'), subtitle=dict(text=dist), yAxis=dict(title=dict(text='发帖量'))))

week3大作业提交/zhangruochen/homework2.py

@@ -0,0 +1,122 @@
+
+# coding: utf-8
+
+# In[2]:
+
+#各类目中成色与均价（全，99，95，9，8，7及以下）
+
+
+# In[1]:
+
+import pymongo
+import charts
+
+
+# In[31]:
+
+client = pymongo.MongoClient('localhost',27017)
+test = client['test']
+homework = test['homework']
+look = test['look']
+
+
+# In[35]:
+
+for i in homework.find():
+    try:
+        if i['price'] == '面议':
+            int_price = False
+        elif u' 元' in i['price']:
+            int_price = int(i['price'][:-2])
+        elif u'元' in i['price']:
+            int_price = int(i['price'][:-1])
+        elif i['price'] == '${info.paramsMap.minprice}':
+            int_price = False
+        else:
+            int_price = int(i['price'])
+    except ValueError:
+        int_price = False
+    homework.update_one({'_id':i['_id']},{'$set':{'int_price':int_price}})
+
+
+# In[50]:
+
+look_list=[]
+for i in homework.find():
+    look_list.append(i['look'])
+look_index_org = set(look_list)
+look_index_org.discard('-')
+look_index_org.discard('${info.paramsMap.oldlevel}')
+look_index_org.discard('报废机/尸体')
+look_index = list(look_index_org)
+look_index.sort()
+look_index.remove('9成新')
+look_index.insert(2,'9成新')
+print(look_index)
+
+
+# In[51]:
+
+counts=[]
+for index in look_index:
+    counts.append(look_list.count(index))
+print(counts)
+
+
+# In[5]:
+
+cates_list = []
+for i in homework.find():
+    cates_list.append(i['cates'])
+cates_index = set(cates_list)
+print(cates_index)
+
+
+# In[74]:
+
+def get_all_price(cates):
+    avg_list=[]
+    for look in look_index:
+        price_list = []
+        for i in homework.find(({'cates':cates,'look':look})):
+            if i['int_price'] != False:
+                price_list.append(i['int_price'])
+            else:
+                pass
+        count = float(len(price_list))
+        sum = 0
+        for i in price_list:
+            sum += i
+        try:
+            avg = sum/count
+            avg_list.append(avg)           
+            data = {
+                'name':'均价',
+                'data':avg_list,
+                'type':'line'
+                }
+        except ZeroDivisionError:
+            pass
+    yield data
+
+
+# In[77]:
+
+for cates in cates_index:
+    options={
+    'chart' : {'zoomType':'xy'},
+    'title' : {'text': '二手%s成色对应的平均价格'%cates},
+    'xAxis' : {'categories': [i for i in look_index]},
+    'yAxis' : {'title': {'text': '价格'}}
+    }
+    series = [i for i in get_all_price(cates)]
+    charts.plot(series, options=options)
+
+
+# In[68]:
+
+def get_details(cates):
+    for look in look_index:
+        for i in homework.find(({'cates':cates,'look':look})):
+            print(i)
+

week3大作业提交/zhangruochen/result1.csv

@@ -0,0 +1,66 @@
+{"_id":{"$oid":"56ac9d5a3bd98f7f15a22c9b"},"area":"����","name":"����/����/����","data":110,"rank":1}
+{"_id":{"$oid":"56ac9d5a3bd98f7f15a22c9c"},"area":"����","name":"�ҵ�","data":66,"rank":2}
+{"_id":{"$oid":"56ac9d5a3bd98f7f15a22c9d"},"area":"����","name":"�����Ʒ","data":40,"rank":3}
+{"_id":{"$oid":"56ac9d753bd98f7f15a22c9e"},"area":"����","name":"�ҵ�","data":3892,"rank":1}
+{"_id":{"$oid":"56ac9d753bd98f7f15a22c9f"},"area":"����","name":"����/����/����","data":1905,"rank":2}
+{"_id":{"$oid":"56ac9d753bd98f7f15a22ca0"},"area":"����","name":"��װ/Ьñ/���","data":1766,"rank":3}
+{"_id":{"$oid":"56ac9d8d3bd98f7f15a22ca1"},"area":"����","name":"����/����/����","data":113,"rank":1}
+{"_id":{"$oid":"56ac9d8d3bd98f7f15a22ca2"},"area":"����","name":"�����Ʒ","data":49,"rank":2}
+{"_id":{"$oid":"56ac9d8d3bd98f7f15a22ca3"},"area":"����","name":"�ҵ�","data":46,"rank":3}
+{"_id":{"$oid":"56ac9da33bd98f7f15a22ca4"},"area":"����","name":"�ҵ�","data":1676,"rank":1}
+{"_id":{"$oid":"56ac9da33bd98f7f15a22ca5"},"area":"����","name":"�Ҿ�","data":455,"rank":2}
+{"_id":{"$oid":"56ac9da33bd98f7f15a22ca6"},"area":"����","name":"����/����/����","data":431,"rank":3}
+{"_id":{"$oid":"56ac9db93bd98f7f15a22ca7"},"area":"�����ܱ�","name":"��װ/Ьñ/���","data":78,"rank":1}
+{"_id":{"$oid":"56ac9db93bd98f7f15a22ca8"},"area":"�����ܱ�","name":"����/����/����","data":71,"rank":2}
+{"_id":{"$oid":"56ac9db93bd98f7f15a22ca9"},"area":"�����ܱ�","name":"�豸","data":64,"rank":3}
+{"_id":{"$oid":"56ac9dcf3bd98f7f15a22caa"},"area":"˳��","name":"�ҵ�","data":386,"rank":1}
+{"_id":{"$oid":"56ac9dcf3bd98f7f15a22cab"},"area":"˳��","name":"����/����/����","data":349,"rank":2}
+{"_id":{"$oid":"56ac9dcf3bd98f7f15a22cac"},"area":"˳��","name":"�칫��Ʒ/�豸","data":147,"rank":3}
+{"_id":{"$oid":"56ac9de43bd98f7f15a22cad"},"area":"ʯ��ɽ","name":"�ҵ�","data":578,"rank":1}
+{"_id":{"$oid":"56ac9de43bd98f7f15a22cae"},"area":"ʯ��ɽ","name":"����/����/����","data":163,"rank":2}
+{"_id":{"$oid":"56ac9de43bd98f7f15a22caf"},"area":"ʯ��ɽ","name":"��װ/Ьñ/���","data":152,"rank":3}
+{"_id":{"$oid":"56ac9dfa3bd98f7f15a22cb0"},"area":"��ͷ��","name":"�ҵ�","data":135,"rank":1}
+{"_id":{"$oid":"56ac9dfa3bd98f7f15a22cb1"},"area":"��ͷ��","name":"�����Ʒ","data":53,"rank":2}
+{"_id":{"$oid":"56ac9dfa3bd98f7f15a22cb2"},"area":"��ͷ��","name":"����/����/����","data":49,"rank":3}
+{"_id":{"$oid":"56ac9e123bd98f7f15a22cb3"},"area":"����","name":"�ҵ�","data":2271,"rank":1}
+{"_id":{"$oid":"56ac9e123bd98f7f15a22cb4"},"area":"����","name":"�����Ʒ","data":1591,"rank":2}
+{"_id":{"$oid":"56ac9e123bd98f7f15a22cb5"},"area":"����","name":"����/����/����","data":945,"rank":3}
+{"_id":{"$oid":"56ac9e283bd98f7f15a22cb6"},"area":"����","name":"�ҵ�","data":378,"rank":1}
+{"_id":{"$oid":"56ac9e283bd98f7f15a22cb7"},"area":"����","name":"��װ/Ьñ/���","data":200,"rank":2}
+{"_id":{"$oid":"56ac9e283bd98f7f15a22cb8"},"area":"����","name":"����/����/����","data":177,"rank":3}
+{"_id":{"$oid":"56ac9e3e3bd98f7f15a22cb9"},"area":"��ɽ","name":"�ҵ�","data":456,"rank":1}
+{"_id":{"$oid":"56ac9e3e3bd98f7f15a22cba"},"area":"��ɽ","name":"����/����/����","data":185,"rank":2}
+{"_id":{"$oid":"56ac9e3e3bd98f7f15a22cbb"},"area":"��ɽ","name":"ĸӤ/��ͯ��Ʒ","data":134,"rank":3}
+{"_id":{"$oid":"56ac9e553bd98f7f15a22cbc"},"area":"��̨","name":"�ҵ�","data":2355,"rank":1}
+{"_id":{"$oid":"56ac9e553bd98f7f15a22cbd"},"area":"��̨","name":"����/����/����","data":838,"rank":2}
+{"_id":{"$oid":"56ac9e553bd98f7f15a22cbe"},"area":"��̨","name":"�Ҿ�","data":651,"rank":3}
+{"_id":{"$oid":"56ac9e6b3bd98f7f15a22cbf"},"area":"����","name":"�Ҿ�","data":3,"rank":1}
+{"_id":{"$oid":"56ac9e6b3bd98f7f15a22cc0"},"area":"����","name":"ƽ�����","data":0,"rank":2}
+{"_id":{"$oid":"56ac9e6b3bd98f7f15a22cc1"},"area":"����","name":"�ֻ�","data":0,"rank":3}
+{"_id":{"$oid":"56ac9e823bd98f7f15a22cc2"},"area":null,"name":"ͼ��/����/����","data":2134,"rank":1}
+{"_id":{"$oid":"56ac9e823bd98f7f15a22cc3"},"area":null,"name":"��װ/Ьñ/���","data":1918,"rank":2}
+{"_id":{"$oid":"56ac9e823bd98f7f15a22cc4"},"area":null,"name":"�����Ʒ","data":1607,"rank":3}
+{"_id":{"$oid":"56ac9e993bd98f7f15a22cc5"},"area":"��ƽ","name":"�ҵ�","data":1129,"rank":1}
+{"_id":{"$oid":"56ac9e993bd98f7f15a22cc6"},"area":"��ƽ","name":"����/����/����","data":460,"rank":2}
+{"_id":{"$oid":"56ac9e993bd98f7f15a22cc7"},"area":"��ƽ","name":"ĸӤ/��ͯ��Ʒ","data":427,"rank":3}
+{"_id":{"$oid":"56ac9eae3bd98f7f15a22cc8"},"area":"�ོ","name":"�ҵ�","data":170,"rank":1}
+{"_id":{"$oid":"56ac9eae3bd98f7f15a22cc9"},"area":"�ོ","name":"����/����/����","data":58,"rank":2}
+{"_id":{"$oid":"56ac9eae3bd98f7f15a22cca"},"area":"�ོ","name":"�����Ʒ","data":48,"rank":3}
+{"_id":{"$oid":"56ac9ec33bd98f7f15a22ccb"},"area":"����","name":"�����Ʒ","data":34,"rank":1}
+{"_id":{"$oid":"56ac9ec33bd98f7f15a22ccc"},"area":"����","name":"����/����/����","data":32,"rank":2}
+{"_id":{"$oid":"56ac9ec33bd98f7f15a22ccd"},"area":"����","name":"ĸӤ/��ͯ��Ʒ","data":18,"rank":3}
+{"_id":{"$oid":"56ac9edb3bd98f7f15a22cce"},"area":"����","name":"�ҵ�","data":560,"rank":1}
+{"_id":{"$oid":"56ac9edb3bd98f7f15a22ccf"},"area":"����","name":"��װ/Ьñ/���","data":558,"rank":2}
+{"_id":{"$oid":"56ac9edb3bd98f7f15a22cd0"},"area":"����","name":"����/����/����","data":354,"rank":3}
+{"_id":{"$oid":"56ac9ef23bd98f7f15a22cd1"},"area":"����","name":"�ҵ�","data":188,"rank":1}
+{"_id":{"$oid":"56ac9ef23bd98f7f15a22cd2"},"area":"����","name":"����/����/����","data":135,"rank":2}
+{"_id":{"$oid":"56ac9ef23bd98f7f15a22cd3"},"area":"����","name":"�칫��Ʒ/�豸","data":114,"rank":3}
+{"_id":{"$oid":"56ac9f083bd98f7f15a22cd4"},"area":"ƽ��","name":"����/����/����","data":74,"rank":1}
+{"_id":{"$oid":"56ac9f083bd98f7f15a22cd5"},"area":"ƽ��","name":"�����Ʒ","data":33,"rank":2}
+{"_id":{"$oid":"56ac9f083bd98f7f15a22cd6"},"area":"ƽ��","name":"̨ʽ��/���","data":21,"rank":3}
+{"_id":{"$oid":"56ac9f1f3bd98f7f15a22cd7"},"area":"����","name":"�ҵ�","data":501,"rank":1}
+{"_id":{"$oid":"56ac9f1f3bd98f7f15a22cd8"},"area":"����","name":"����/����/����","data":382,"rank":2}
+{"_id":{"$oid":"56ac9f1f3bd98f7f15a22cd9"},"area":"����","name":"��װ/Ьñ/���","data":330,"rank":3}
+{"_id":{"$oid":"56ac9f373bd98f7f15a22cda"},"area":"ͨ��","name":"�ҵ�","data":1149,"rank":1}
+{"_id":{"$oid":"56ac9f373bd98f7f15a22cdb"},"area":"ͨ��","name":"����/����/����","data":638,"rank":2}
+{"_id":{"$oid":"56ac9f373bd98f7f15a22cdc"},"area":"ͨ��","name":"ĸӤ/��ͯ��Ʒ","data":389,"rank":3}