I have a list of type:
ans=[(a,[b,c]),(x,[y,z]),(p,[q,r])]
I need to sort the list by using the following condition :
if (ans[j][1][1]>ans[j+1][1][1]) or (ans[j][1][1]==ans[j+1][1][1] and ans[j][1][0]<ans[j+1][1][0]):
# do something (like swap(ans[j],ans[j+1]))
I was able to implement using bubble sort, but I want a faster sorting method.
Is there a way to sort my list using the sort() or sorted() (Using comparator or something similar) functions while pertaining to my condition ?
You can create a comparator function that retuns a tuple; tuples are compared from left to right until one of the elements is "larger" than the other. Your input/output example is quite lacking, but I believe this will result into what you want:
def my_compare(x):
return x[1][1], x[1][0]
ans.sort(key=my_compare)
# ans = sorted(ans, key=my_compare)
Essentially this will first compare the x[1][1] value of both ans[j] and ans[j+1], and if it's the same then it will compare the x[1][0] value. You can rearrange and add more comparators as you wish if this didn't match your ues case perfectly.
Related
I used a list of a constraint numbers or a list of constraint names but this does not work with command m.remove_constraints().
m.add(sumbs[i] <= a[i], "ct_sum_%d" %i)
A.append("ct_sum_%d" %i)
Then later on when I want to change the model: m.remove_constraints(A)
What is the correct way of doing this?
Model.add_constraints() returns the list of newly added constraints.
Similarly, Model.remove_constraints takes a collection of constraint objects, not names , not indices. For example:
cts = mdl.add_constraints(...)
mdl.remove_constraints(cts[:3]) # remove the first three
I would appreciate suggestions for a more computationally efficient way to dynamically filter a Pandas DataFrame.
The size of the DataFrame, len(df.index), is around 680,000.
This code from the callback function of a Plotly Dash dashboard is triggered when points on a scatter graph are selected. These points are passed to points as a list of dictionaries containing various properties with keys 'A' to 'C'. This allows the user to select a subset of the data in the pandas.DataFrame instance df for cross-filtering analysis.
rows_boolean = pandas.Series([False] * len(df.index))
for point in points:
current_condition = ((df['A'] == point['a']) & (df['B'] == point['b'])
& (df['C'] >= point['c']) & (df['C'] < point['d']))
rows_boolean = rows_boolean | current_condition
filtered = df.loc[rows_boolean, list_of_column_names]
The body of this for loop is very slow as it is iterating over the whole data frame, it is manageable to run it once but not inside a loop.
Note that these filters are not additive, as in this example; each successive iteration of the for loop increases, rather than decreases, the size of filtered (as | rather than & operator is used).
Note also that I am aware of the existence of the method df['C'].between(point['c'], point['d']) as an alternative to the last two comparison operators, however, I only want this comparison to be inclusive at the lower end.
Solutions I have considered
Searching the many frustratingly similar posts on SO reveals a few ideas which get some of the way:
Using pandas.DataFrame.query() will require building a (potentially very large) query string as follows:
query = ' | '.join([f'((A == {point["a"]}) & (B == {point["b"]})
& (C >= {point["c"]}) & (C < {point["d"]}))' for point in points])
filtered = df.query(query)
My main concern here is that I don’t know how efficient the query method becomes when the query passed has several dozen (or even several hundred) conditions strung together. This solution also currently does not allow the selection of columns using list_of_column_names.
Another possible solution could come from implementing something like this.
To reiterate, speed is key here, so I'm not just after something that works, but something that works a darn sight faster than my boolean implementation above:
There should be one-- and preferably only one --obvious way to do it. (PEP 20)
I'm doing the following traversal:
g.V().has('Transfer','eventName','Airdrop').as('t1').
outE('sent_to').
inV().dedup().as('a2').
inE('sent_from').
outV().as('t2').
where('t1',eq('t2')).by('address').
outE('sent_to').
inV().as('a3').
select('a3','a2').
by('accountId').toList().groupBy { it.a3 }.collectEntries { [(it.key): [a2 : it.value.a2]]};
So as you can see I'm basically doing a traversal and at the end I'm using groovy with collectEntries to aggregate the results like I need them, which is aggregated by a3 in this case. The results look like this:
==>0xfe43502662ce2adf86d9d49f25a27d65c70a709d={a2=[0x99feb505a8ed9976cf19e757a9536117e6cdc5ba, 0x22019ad32ea3adabae68003bdefd099d7e5e3886]}
(This is GOOD, because the number of values in a2 is at least 2)
==>0x129e0131ea3cc16fe5252d7280bd1258f629f20f={a2=[0xf7958fad496d15cf9fd9e54c0012504f4fdb96ff]}
(This is NOT GOOD, I want to return in my list only those combinations where there are at lest 2 values for a2)
I have tried using filters and an additional where step in the traversal itself but I haven't been able to do it. I'm not sure if this is something I should skip using Groovy in my last line. Any help or orientation would be very much appreciated
I don't think you need to drop into Groovy to get the answer you want. It would be preferable to do this all in Gremlin especially since you intend to filter results which could yield some performance benefit. Gremlin has it's own group() step as well as methods for filtering the resulting Map:
g.V().has('Transfer','eventName','Airdrop').as('t1').
out('sent_to').
dedup().as('a2').
in('sent_from').as('t2').
where('t1',eq('t2')).by('address').
out('sent_to').inV().as('a3').
select('a3','a2').
by('accountId').
group().
by('a3').
by('a2').
unfold().
where(select(values).limit(local,2).count(local).is(gte(2)))
The idea is to build your Map with group() then deconstruct it to entries with unfold(). You the filter each entry with where() by selecting the values of the entry, which is a List of "a2" then counting the items locally in that List. I use limit(local,2) to avoid unnecessary iteration beyond 2 since the filter is gte(2).
The easiest way to do this is with findAll { }.
.groupBy { it.a3 }
.findAll { it.value.a2.size() > 1 }
.collectEntries { [(it.key): [a2: it.value.a2]] }
if some a2 are null, then value.a2 also evaluates to null and filters the results without the need for explicit nullchecks
What I want to do is trying to get different values of p_y_given_x for different sigma2_n separately in each line? When I use append it gives me result in one array. But I want separate results for each 0.2 for p_y_given_x.
Is there any hint for me?
for sigma2_n in np.arange(0.2,0.9):
p_y_given_x= np.exp(-(y_new-alphabet[t])**2/2/sigma2_N)
the result should be like this
p_y_given_x1=
p_y_given_x2=
p_y_given_x3=
....
ummmmm.... actually maybe this helps f(x)=x+2, f(1)=1+2, f(2)=2+2 and etc, I want to use a for loop and define x inside it and for each iteration use one values of x and insert it inside the function and calculate the f(x), but i would like to get values of f(x) separately and compare them, for example like this:
f(1)
f(2)
f(3)(separately in different lines)
not like this in one array[f(1),f(2),f(3)] for the last case we can use append()
I have a query calculation that should throw me either a value (if conditions are met) or a blank/null value.
The code is in the following form:
if([attribute] > 3)
then ('value')
else ('')
At the moment the only way I could find to obtain the result is the use of '' (i.e. an empty character string), but this a value as well, so when I subsequently count the number of distinct values in another query I struggle to get the correct number (the empty string should be removed from the count, if found).
I can get the result with the following code:
if (attribute='') in ([first_query].[attribute]))
then (count(distinct(attribute)-1)
else (count(distinct(attribute))
How to avoid the double calculation in all later queries involving the count of attribute?
I use this Cognos function:
nullif(1, 1)
I found out that this can be managed using the case when function:
case
when ([attribute] > 3)
then ('value')
end
The difference is that case when doesn't need to have all the possible options for Handling data, and if it founds a case that is not in the list it just returns a blank cell.
Perfect for what I needed (and not as well documented on the web as the opposite case, i.e. dealing with null cases that should be zero).