Update 45 2022 0614 heap 8th.md

Author: the-yuyut

紀錄/45 2022 0614 heap 8th.md

@@ -12,7 +12,6 @@
 8. 認真在考慮之後來看面試綠皮書
 
 
---- 
 
 * 	[done] MEDIUM	215	Kth Largest Element in an Array	https://leetcode.com/problems/kth-largest-element-in-an-array
 * 	[done] MEDIUM	347	Top K Frequent Elements	https://leetcode.com/problems/top-k-frequent-elements
@@ -24,8 +23,93 @@
 * 	[done] MEDIUM	973	K Closest Points to Origin	https://leetcode.com/problems/k-closest-points-to-origin
 * 	🔓	MEDIUM	1167	Minimum Cost to Connect Sticks	https://leetcode.com/problems/minimum-cost-to-connect-sticks
 * 	[done] MEDIUM	1642	Furthest Building You Can Reach	https://leetcode.com/problems/furthest-building-you-can-reach
-* 	HARD	295	Find Median from Data Stream	https://leetcode.com/problems/find-median-from-data-stream
+* 	[done] HARD	295	Find Median from Data Stream	https://leetcode.com/problems/find-median-from-data-stream
 *   MEDIUM 1631 Path With Minimum Effort https://leetcode.com/problems/path-with-minimum-effort/
 
 ---
 
+```python
+"""
+295
+"""
+from heapq import *        # In python the heap is a MIN heap
+
+class MedianFinder:
+    def __init__(self):
+        self.small = []             # Stores the number that smaller than median, a max heap (* -1)
+        self.large = []             # Stores the number that larger than median, a min heap
+        self.size = 0               # Stores the size of data
+
+    def addNum(self, num):
+        heappush(self.small, (-1)*heappushpop(self.large, num))
+        self.size += 1
+
+        # Adjust the size of both heap,
+        while len(self.large) < len(self.small):        # keep the large always slightly bigger
+            heappush(self.large, (-1)*heappop(self.small))
+
+    def findMedian(self):
+        if self.size%2:
+            return float(self.large[0])
+        else:
+            return float(self.large[0] + (-1)*self.small[0])/2.0
+```
+--- 
+
+cat ... | do...
+
+curl | ...
+stdinput
+
+
+Y.J. Lee下午8:02
+XD
+Y.J. Lee下午8:07
+Python就是之前ML很紅的時候很夯
+Y.J. Lee下午8:08
+有滿多免費的ML的demo專案
+幾乎一定會碰到visualization的東西
+不難啦 都是直接用library
+sry, 我好像每次都直接貼中間 qq
+Y.J. Lee下午8:17
+一樣的  ^^
+無腦調整 size
+Y.J. Lee下午8:19
+add的時候先上鎖 加完在解鎖
+Y.J. Lee下午8:21
+嗯 合理
+Y.J. Lee下午8:26
+嗯 他會動態調整memory
+Y.J. Lee下午8:33
+我們是用這個
+https://aws.amazon.com/kinesis/data-firehose/
+接好以後DB 的更新會自己propagate 到date warehouse
+data warehouse
+Y.J. Lee下午8:40
+那邊直接push也是可以啦 不過就while那邊要改一下
+Y.J. Lee下午8:42
+這題另外兩個做法就是 binary search tree 跟 sorted list 不確定swift有沒有library
+python sortedlist 可以 O(logN)加東西 O(1)取index
+他實作滿複雜的
+好像真的是很多個小list
+Y.J. Lee下午8:43
+對 不管怎麼加東西 他自己都是維持sorted
+bst要balance
+Y.J. Lee下午8:45
+對啊 就很複雜 不見得比較好
+Y.J. Lee下午8:47
+用不到 但是面試可能會考XD
+Y.J. Lee下午8:50
+類似 bucket sort
+都變 O(1)
+Y.J. Lee下午8:52
+固定 number of buckets = 100, O(100) 也算是常數
+Y.J. Lee下午8:56
+sliding window median其實真的很類似
+就maintain 一個 toRemove (set?)
+window 裡面一樣兩個heap
+我猜的 也沒寫過 XD
+Y.J. Lee下午9:01
+QQ
+Y.J. Lee下午9:02
+先撤拉 晚安~