759. Employee Free Time
Problem Explanation:
Given a list of employee work schedules with each employee having a list of non-overlapping intervals representing their working hours, we are tasked with finding the common free time for all employees, or in other words, the times when all employees are not working.
The input is a nested list of intervals, each interval as [start, end]
, with start < end
. The intervals are non-overlapping and are already sorted in ascending order. The output should also be a list of sorted intervals.
For example, consider schedule = [[[1,3],[6,7]],[[2,4]],[[2,5],[9,12]]]
. Here, Employee 1 works from 1 to 3 and 6 to 7. Employee 2 works from 2 to 4 and Employee 3 works from 2 to 5 and 9 to 12. The common free time for all employees is [5,6]
and [7,9]
as these are the intervals when all employees are free.
Solution Approach:
The proposed solution concatenates all work schedules into a single list, sorts this list and then identifies the gaps between work intervals, which represent the common free time for all employees.
First, the solution creates an empty array to store the result. Then, it goes through the work shifts of each employee and combines all the intervals to a single list.
Next, this combined list of work intervals is sorted using the start of the work intervals. This will ensure that the work schedules are in chronological order.
After that, the solution stores the end time of the first shift in a variable called prevEnd
, and then iterates through each work interval.
If the start of a work interval is greater than prevEnd
, this means there's a gap between the end of the previous work shift and the start of the current work shift. This gap represents a common free time, which is then added to the result list.
Finally, prevEnd
is updated with the maximum end time of the current work interval or the prevEnd
.
Through this approach, the algorithm successfully identifies the common free time of all employees.
Python Solution
python # Definition for an Interval. # class Interval: # def __init__(self, start: int = None, end: int = None): # self.start = start # self.end = end class Solution: def employeeFreeTime(self, schedule: '[[Interval]]') -> '[Interval]': # Flattening the schedule intervals = [interval for employee in schedule for interval in employee] # Sorting by start of each Interval intervals.sort(key=lambda x: x.start) res, end = [], intervals[0].end # Checking for free time between intervals for i in intervals[1:]: if end < i.start: res.append(Interval(end, i.start)) end = max(end, i.end) return res
JavaScript Solution:
javascript
function Interval(start, end) {
this.start = start;
this.end = end;
}
function employeeFreeTime(schedule) {
// Flattening the schedule
let intervals = [].concat(...schedule);
// Sorting by start of each Interval
intervals.sort((a, b) => a.start - b.start);
let res = [], end = intervals[0].end;
// Checking for free time between intervals
for (let i = 1; i < intervals.length; i++) {
if (end < intervals[i].start) {
res.push(new Interval(end, intervals[i].start));
}
end = Math.max(end, intervals[i].end);
}
return res;
}
Java Solution:
java
/**
* Definition for an interval.
* public class Interval {
* int start;
* int end;
* Interval() { start = 0; end = 0; }
* Interval(int s, int e) { start = s; end = e; }
* }
*/
public class Solution {
public List<Interval> employeeFreeTime(List<List<Interval>> schedule) {
List<Interval> res = new ArrayList<>();
List<Interval> intervals = new ArrayList<>();
// Flattening the schedule
for (List<Interval> employee : schedule)
for (Interval interval : employee)
intervals.add(interval);
// Sorting by start of each Interval
Collections.sort(intervals, (a, b) -> a.start - b.start);
int end = intervals.get(0).end;
// Checking for free time between intervals
for (Interval interval : intervals) {
if (interval.start > end)
res.add(new Interval(end, interval.start));
end = Math.max(end, interval.end);
}
return res;
}
}
These solutions work by first flattening the schedule list, then sorting the intervals by their start times. They then keep track of the end time of the current interval. If the start of the next interval is greater than this end time, it means there is a gap where all employees are free, and this gap is then added to the results. These solutions have a time complexity of O(N log N) due to the sorting operation, where N is the total number of intervals. The space complexity is also O(N) for storing the intervals and the result.
Ready to land your dream job?
Unlock your dream job with a 2-minute evaluator for a personalized learning plan!
Start EvaluatorGiven a sorted array of integers and an integer called target, find the element that
equals to the target and return its index. Select the correct code that fills the
___
in the given code snippet.
1def binary_search(arr, target):
2 left, right = 0, len(arr) - 1
3 while left ___ right:
4 mid = (left + right) // 2
5 if arr[mid] == target:
6 return mid
7 if arr[mid] < target:
8 ___ = mid + 1
9 else:
10 ___ = mid - 1
11 return -1
12
1public static int binarySearch(int[] arr, int target) {
2 int left = 0;
3 int right = arr.length - 1;
4
5 while (left ___ right) {
6 int mid = left + (right - left) / 2;
7 if (arr[mid] == target) return mid;
8 if (arr[mid] < target) {
9 ___ = mid + 1;
10 } else {
11 ___ = mid - 1;
12 }
13 }
14 return -1;
15}
16
1function binarySearch(arr, target) {
2 let left = 0;
3 let right = arr.length - 1;
4
5 while (left ___ right) {
6 let mid = left + Math.trunc((right - left) / 2);
7 if (arr[mid] == target) return mid;
8 if (arr[mid] < target) {
9 ___ = mid + 1;
10 } else {
11 ___ = mid - 1;
12 }
13 }
14 return -1;
15}
16
Recommended Readings
LeetCode Patterns Your Personal Dijkstra's Algorithm to Landing Your Dream Job The goal of AlgoMonster is to help you get a job in the shortest amount of time possible in a data driven way We compiled datasets of tech interview problems and broke them down by patterns This way we
Recursion Recursion is one of the most important concepts in computer science Simply speaking recursion is the process of a function calling itself Using a real life analogy imagine a scenario where you invite your friends to lunch https algomonster s3 us east 2 amazonaws com recursion jpg You first
Runtime Overview When learning about algorithms and data structures you'll frequently encounter the term time complexity This concept is fundamental in computer science and offers insights into how long an algorithm takes to complete given a certain input size What is Time Complexity Time complexity represents the amount of time
Want a Structured Path to Master System Design Too? Don’t Miss This!